Actual source code: mpisbaij.c
1: #include <../src/mat/impls/baij/mpi/mpibaij.h>
2: #include <../src/mat/impls/sbaij/mpi/mpisbaij.h>
3: #include <../src/mat/impls/sbaij/seq/sbaij.h>
4: #include <petscblaslapack.h>
6: static PetscErrorCode MatDestroy_MPISBAIJ(Mat mat)
7: {
8: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ *)mat->data;
10: PetscFunctionBegin;
11: PetscCall(PetscLogObjectState((PetscObject)mat, "Rows=%" PetscInt_FMT ",Cols=%" PetscInt_FMT, mat->rmap->N, mat->cmap->N));
12: PetscCall(MatStashDestroy_Private(&mat->stash));
13: PetscCall(MatStashDestroy_Private(&mat->bstash));
14: PetscCall(MatDestroy(&baij->A));
15: PetscCall(MatDestroy(&baij->B));
16: #if defined(PETSC_USE_CTABLE)
17: PetscCall(PetscHMapIDestroy(&baij->colmap));
18: #else
19: PetscCall(PetscFree(baij->colmap));
20: #endif
21: PetscCall(PetscFree(baij->garray));
22: PetscCall(VecDestroy(&baij->lvec));
23: PetscCall(VecScatterDestroy(&baij->Mvctx));
24: PetscCall(VecDestroy(&baij->slvec0));
25: PetscCall(VecDestroy(&baij->slvec0b));
26: PetscCall(VecDestroy(&baij->slvec1));
27: PetscCall(VecDestroy(&baij->slvec1a));
28: PetscCall(VecDestroy(&baij->slvec1b));
29: PetscCall(VecScatterDestroy(&baij->sMvctx));
30: PetscCall(PetscFree2(baij->rowvalues, baij->rowindices));
31: PetscCall(PetscFree(baij->barray));
32: PetscCall(PetscFree(baij->hd));
33: PetscCall(VecDestroy(&baij->diag));
34: PetscCall(VecDestroy(&baij->bb1));
35: PetscCall(VecDestroy(&baij->xx1));
36: #if defined(PETSC_USE_REAL_MAT_SINGLE)
37: PetscCall(PetscFree(baij->setvaluescopy));
38: #endif
39: PetscCall(PetscFree(baij->in_loc));
40: PetscCall(PetscFree(baij->v_loc));
41: PetscCall(PetscFree(baij->rangebs));
42: PetscCall(PetscFree(mat->data));
44: PetscCall(PetscObjectChangeTypeName((PetscObject)mat, NULL));
45: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatStoreValues_C", NULL));
46: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatRetrieveValues_C", NULL));
47: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatMPISBAIJSetPreallocation_C", NULL));
48: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatMPISBAIJSetPreallocationCSR_C", NULL));
49: #if defined(PETSC_HAVE_ELEMENTAL)
50: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatConvert_mpisbaij_elemental_C", NULL));
51: #endif
52: #if defined(PETSC_HAVE_SCALAPACK)
53: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatConvert_mpisbaij_scalapack_C", NULL));
54: #endif
55: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatConvert_mpisbaij_mpiaij_C", NULL));
56: PetscCall(PetscObjectComposeFunction((PetscObject)mat, "MatConvert_mpisbaij_mpibaij_C", NULL));
57: PetscFunctionReturn(PETSC_SUCCESS);
58: }
60: /* defines MatSetValues_MPI_Hash(), MatAssemblyBegin_MPI_Hash(), MatAssemblyEnd_MPI_Hash(), MatSetUp_MPI_Hash() */
61: #define TYPE SBAIJ
62: #define TYPE_SBAIJ
63: #include "../src/mat/impls/aij/mpi/mpihashmat.h"
64: #undef TYPE
65: #undef TYPE_SBAIJ
67: #if defined(PETSC_HAVE_ELEMENTAL)
68: PETSC_INTERN PetscErrorCode MatConvert_MPISBAIJ_Elemental(Mat, MatType, MatReuse, Mat *);
69: #endif
70: #if defined(PETSC_HAVE_SCALAPACK)
71: PETSC_INTERN PetscErrorCode MatConvert_SBAIJ_ScaLAPACK(Mat, MatType, MatReuse, Mat *);
72: #endif
74: /* This could be moved to matimpl.h */
75: static PetscErrorCode MatPreallocateWithMats_Private(Mat B, PetscInt nm, Mat X[], PetscBool symm[], PetscBool fill)
76: {
77: Mat preallocator;
78: PetscInt r, rstart, rend;
79: PetscInt bs, i, m, n, M, N;
80: PetscBool cong = PETSC_TRUE;
82: PetscFunctionBegin;
85: for (i = 0; i < nm; i++) {
87: PetscCall(PetscLayoutCompare(B->rmap, X[i]->rmap, &cong));
88: PetscCheck(cong, PetscObjectComm((PetscObject)B), PETSC_ERR_SUP, "Not for different layouts");
89: }
91: PetscCall(MatGetBlockSize(B, &bs));
92: PetscCall(MatGetSize(B, &M, &N));
93: PetscCall(MatGetLocalSize(B, &m, &n));
94: PetscCall(MatCreate(PetscObjectComm((PetscObject)B), &preallocator));
95: PetscCall(MatSetType(preallocator, MATPREALLOCATOR));
96: PetscCall(MatSetBlockSize(preallocator, bs));
97: PetscCall(MatSetSizes(preallocator, m, n, M, N));
98: PetscCall(MatSetUp(preallocator));
99: PetscCall(MatGetOwnershipRange(preallocator, &rstart, &rend));
100: for (r = rstart; r < rend; ++r) {
101: PetscInt ncols;
102: const PetscInt *row;
103: const PetscScalar *vals;
105: for (i = 0; i < nm; i++) {
106: PetscCall(MatGetRow(X[i], r, &ncols, &row, &vals));
107: PetscCall(MatSetValues(preallocator, 1, &r, ncols, row, vals, INSERT_VALUES));
108: if (symm && symm[i]) PetscCall(MatSetValues(preallocator, ncols, row, 1, &r, vals, INSERT_VALUES));
109: PetscCall(MatRestoreRow(X[i], r, &ncols, &row, &vals));
110: }
111: }
112: PetscCall(MatAssemblyBegin(preallocator, MAT_FINAL_ASSEMBLY));
113: PetscCall(MatAssemblyEnd(preallocator, MAT_FINAL_ASSEMBLY));
114: PetscCall(MatPreallocatorPreallocate(preallocator, fill, B));
115: PetscCall(MatDestroy(&preallocator));
116: PetscFunctionReturn(PETSC_SUCCESS);
117: }
119: PETSC_INTERN PetscErrorCode MatConvert_MPISBAIJ_Basic(Mat A, MatType newtype, MatReuse reuse, Mat *newmat)
120: {
121: Mat B;
122: PetscInt r;
124: PetscFunctionBegin;
125: if (reuse != MAT_REUSE_MATRIX) {
126: PetscBool symm = PETSC_TRUE, isdense;
127: PetscInt bs;
129: PetscCall(MatCreate(PetscObjectComm((PetscObject)A), &B));
130: PetscCall(MatSetSizes(B, A->rmap->n, A->cmap->n, A->rmap->N, A->cmap->N));
131: PetscCall(MatSetType(B, newtype));
132: PetscCall(MatGetBlockSize(A, &bs));
133: PetscCall(MatSetBlockSize(B, bs));
134: PetscCall(PetscLayoutSetUp(B->rmap));
135: PetscCall(PetscLayoutSetUp(B->cmap));
136: PetscCall(PetscObjectTypeCompareAny((PetscObject)B, &isdense, MATSEQDENSE, MATMPIDENSE, MATSEQDENSECUDA, ""));
137: if (!isdense) {
138: PetscCall(MatGetRowUpperTriangular(A));
139: PetscCall(MatPreallocateWithMats_Private(B, 1, &A, &symm, PETSC_TRUE));
140: PetscCall(MatRestoreRowUpperTriangular(A));
141: } else {
142: PetscCall(MatSetUp(B));
143: }
144: } else {
145: B = *newmat;
146: PetscCall(MatZeroEntries(B));
147: }
149: PetscCall(MatGetRowUpperTriangular(A));
150: for (r = A->rmap->rstart; r < A->rmap->rend; r++) {
151: PetscInt ncols;
152: const PetscInt *row;
153: const PetscScalar *vals;
155: PetscCall(MatGetRow(A, r, &ncols, &row, &vals));
156: PetscCall(MatSetValues(B, 1, &r, ncols, row, vals, INSERT_VALUES));
157: #if defined(PETSC_USE_COMPLEX)
158: if (A->hermitian == PETSC_BOOL3_TRUE) {
159: PetscInt i;
160: for (i = 0; i < ncols; i++) PetscCall(MatSetValue(B, row[i], r, PetscConj(vals[i]), INSERT_VALUES));
161: } else {
162: PetscCall(MatSetValues(B, ncols, row, 1, &r, vals, INSERT_VALUES));
163: }
164: #else
165: PetscCall(MatSetValues(B, ncols, row, 1, &r, vals, INSERT_VALUES));
166: #endif
167: PetscCall(MatRestoreRow(A, r, &ncols, &row, &vals));
168: }
169: PetscCall(MatRestoreRowUpperTriangular(A));
170: PetscCall(MatAssemblyBegin(B, MAT_FINAL_ASSEMBLY));
171: PetscCall(MatAssemblyEnd(B, MAT_FINAL_ASSEMBLY));
173: if (reuse == MAT_INPLACE_MATRIX) {
174: PetscCall(MatHeaderReplace(A, &B));
175: } else {
176: *newmat = B;
177: }
178: PetscFunctionReturn(PETSC_SUCCESS);
179: }
181: static PetscErrorCode MatStoreValues_MPISBAIJ(Mat mat)
182: {
183: Mat_MPISBAIJ *aij = (Mat_MPISBAIJ *)mat->data;
185: PetscFunctionBegin;
186: PetscCall(MatStoreValues(aij->A));
187: PetscCall(MatStoreValues(aij->B));
188: PetscFunctionReturn(PETSC_SUCCESS);
189: }
191: static PetscErrorCode MatRetrieveValues_MPISBAIJ(Mat mat)
192: {
193: Mat_MPISBAIJ *aij = (Mat_MPISBAIJ *)mat->data;
195: PetscFunctionBegin;
196: PetscCall(MatRetrieveValues(aij->A));
197: PetscCall(MatRetrieveValues(aij->B));
198: PetscFunctionReturn(PETSC_SUCCESS);
199: }
201: #define MatSetValues_SeqSBAIJ_A_Private(row, col, value, addv, orow, ocol) \
202: do { \
203: brow = row / bs; \
204: rp = aj + ai[brow]; \
205: ap = aa + bs2 * ai[brow]; \
206: rmax = aimax[brow]; \
207: nrow = ailen[brow]; \
208: bcol = col / bs; \
209: ridx = row % bs; \
210: cidx = col % bs; \
211: low = 0; \
212: high = nrow; \
213: while (high - low > 3) { \
214: t = (low + high) / 2; \
215: if (rp[t] > bcol) high = t; \
216: else low = t; \
217: } \
218: for (_i = low; _i < high; _i++) { \
219: if (rp[_i] > bcol) break; \
220: if (rp[_i] == bcol) { \
221: bap = ap + bs2 * _i + bs * cidx + ridx; \
222: if (addv == ADD_VALUES) *bap += value; \
223: else *bap = value; \
224: goto a_noinsert; \
225: } \
226: } \
227: if (a->nonew == 1) goto a_noinsert; \
228: PetscCheck(a->nonew != -1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Inserting a new nonzero at global row/column (%" PetscInt_FMT ", %" PetscInt_FMT ") into matrix", orow, ocol); \
229: MatSeqXAIJReallocateAIJ(A, a->mbs, bs2, nrow, brow, bcol, rmax, aa, ai, aj, rp, ap, aimax, a->nonew, MatScalar); \
230: N = nrow++ - 1; \
231: /* shift up all the later entries in this row */ \
232: PetscCall(PetscArraymove(rp + _i + 1, rp + _i, N - _i + 1)); \
233: PetscCall(PetscArraymove(ap + bs2 * (_i + 1), ap + bs2 * _i, bs2 * (N - _i + 1))); \
234: PetscCall(PetscArrayzero(ap + bs2 * _i, bs2)); \
235: rp[_i] = bcol; \
236: ap[bs2 * _i + bs * cidx + ridx] = value; \
237: a_noinsert:; \
238: ailen[brow] = nrow; \
239: } while (0)
241: #define MatSetValues_SeqSBAIJ_B_Private(row, col, value, addv, orow, ocol) \
242: do { \
243: brow = row / bs; \
244: rp = bj + bi[brow]; \
245: ap = ba + bs2 * bi[brow]; \
246: rmax = bimax[brow]; \
247: nrow = bilen[brow]; \
248: bcol = col / bs; \
249: ridx = row % bs; \
250: cidx = col % bs; \
251: low = 0; \
252: high = nrow; \
253: while (high - low > 3) { \
254: t = (low + high) / 2; \
255: if (rp[t] > bcol) high = t; \
256: else low = t; \
257: } \
258: for (_i = low; _i < high; _i++) { \
259: if (rp[_i] > bcol) break; \
260: if (rp[_i] == bcol) { \
261: bap = ap + bs2 * _i + bs * cidx + ridx; \
262: if (addv == ADD_VALUES) *bap += value; \
263: else *bap = value; \
264: goto b_noinsert; \
265: } \
266: } \
267: if (b->nonew == 1) goto b_noinsert; \
268: PetscCheck(b->nonew != -1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Inserting a new nonzero at global row/column (%" PetscInt_FMT ", %" PetscInt_FMT ") into matrix", orow, ocol); \
269: MatSeqXAIJReallocateAIJ(B, b->mbs, bs2, nrow, brow, bcol, rmax, ba, bi, bj, rp, ap, bimax, b->nonew, MatScalar); \
270: N = nrow++ - 1; \
271: /* shift up all the later entries in this row */ \
272: PetscCall(PetscArraymove(rp + _i + 1, rp + _i, N - _i + 1)); \
273: PetscCall(PetscArraymove(ap + bs2 * (_i + 1), ap + bs2 * _i, bs2 * (N - _i + 1))); \
274: PetscCall(PetscArrayzero(ap + bs2 * _i, bs2)); \
275: rp[_i] = bcol; \
276: ap[bs2 * _i + bs * cidx + ridx] = value; \
277: b_noinsert:; \
278: bilen[brow] = nrow; \
279: } while (0)
281: /* Only add/insert a(i,j) with i<=j (blocks).
282: Any a(i,j) with i>j input by user is ignored or generates an error
283: */
284: static PetscErrorCode MatSetValues_MPISBAIJ(Mat mat, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], const PetscScalar v[], InsertMode addv)
285: {
286: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ *)mat->data;
287: MatScalar value;
288: PetscBool roworiented = baij->roworiented;
289: PetscInt i, j, row, col;
290: PetscInt rstart_orig = mat->rmap->rstart;
291: PetscInt rend_orig = mat->rmap->rend, cstart_orig = mat->cmap->rstart;
292: PetscInt cend_orig = mat->cmap->rend, bs = mat->rmap->bs;
294: /* Some Variables required in the macro */
295: Mat A = baij->A;
296: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
297: PetscInt *aimax = a->imax, *ai = a->i, *ailen = a->ilen, *aj = a->j;
298: MatScalar *aa = a->a;
300: Mat B = baij->B;
301: Mat_SeqBAIJ *b = (Mat_SeqBAIJ *)B->data;
302: PetscInt *bimax = b->imax, *bi = b->i, *bilen = b->ilen, *bj = b->j;
303: MatScalar *ba = b->a;
305: PetscInt *rp, ii, nrow, _i, rmax, N, brow, bcol;
306: PetscInt low, high, t, ridx, cidx, bs2 = a->bs2;
307: MatScalar *ap, *bap;
309: /* for stash */
310: PetscInt n_loc, *in_loc = NULL;
311: MatScalar *v_loc = NULL;
313: PetscFunctionBegin;
314: if (!baij->donotstash) {
315: if (n > baij->n_loc) {
316: PetscCall(PetscFree(baij->in_loc));
317: PetscCall(PetscFree(baij->v_loc));
318: PetscCall(PetscMalloc1(n, &baij->in_loc));
319: PetscCall(PetscMalloc1(n, &baij->v_loc));
321: baij->n_loc = n;
322: }
323: in_loc = baij->in_loc;
324: v_loc = baij->v_loc;
325: }
327: for (i = 0; i < m; i++) {
328: if (im[i] < 0) continue;
329: PetscCheck(im[i] < mat->rmap->N, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Row too large: row %" PetscInt_FMT " max %" PetscInt_FMT, im[i], mat->rmap->N - 1);
330: if (im[i] >= rstart_orig && im[i] < rend_orig) { /* this processor entry */
331: row = im[i] - rstart_orig; /* local row index */
332: for (j = 0; j < n; j++) {
333: if (im[i] / bs > in[j] / bs) {
334: if (a->ignore_ltriangular) {
335: continue; /* ignore lower triangular blocks */
336: } else SETERRQ(PETSC_COMM_SELF, PETSC_ERR_USER, "Lower triangular value cannot be set for sbaij format. Ignoring these values, run with -mat_ignore_lower_triangular or call MatSetOption(mat,MAT_IGNORE_LOWER_TRIANGULAR,PETSC_TRUE)");
337: }
338: if (in[j] >= cstart_orig && in[j] < cend_orig) { /* diag entry (A) */
339: col = in[j] - cstart_orig; /* local col index */
340: brow = row / bs;
341: bcol = col / bs;
342: if (brow > bcol) continue; /* ignore lower triangular blocks of A */
343: if (roworiented) value = v[i * n + j];
344: else value = v[i + j * m];
345: MatSetValues_SeqSBAIJ_A_Private(row, col, value, addv, im[i], in[j]);
346: /* PetscCall(MatSetValues_SeqBAIJ(baij->A,1,&row,1,&col,&value,addv)); */
347: } else if (in[j] < 0) {
348: continue;
349: } else {
350: PetscCheck(in[j] < mat->cmap->N, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Column too large: col %" PetscInt_FMT " max %" PetscInt_FMT, in[j], mat->cmap->N - 1);
351: /* off-diag entry (B) */
352: if (mat->was_assembled) {
353: if (!baij->colmap) PetscCall(MatCreateColmap_MPIBAIJ_Private(mat));
354: #if defined(PETSC_USE_CTABLE)
355: PetscCall(PetscHMapIGetWithDefault(baij->colmap, in[j] / bs + 1, 0, &col));
356: col = col - 1;
357: #else
358: col = baij->colmap[in[j] / bs] - 1;
359: #endif
360: if (col < 0 && !((Mat_SeqSBAIJ *)baij->A->data)->nonew) {
361: PetscCall(MatDisAssemble_MPISBAIJ(mat));
362: col = in[j];
363: /* Reinitialize the variables required by MatSetValues_SeqBAIJ_B_Private() */
364: B = baij->B;
365: b = (Mat_SeqBAIJ *)B->data;
366: bimax = b->imax;
367: bi = b->i;
368: bilen = b->ilen;
369: bj = b->j;
370: ba = b->a;
371: } else col += in[j] % bs;
372: } else col = in[j];
373: if (roworiented) value = v[i * n + j];
374: else value = v[i + j * m];
375: MatSetValues_SeqSBAIJ_B_Private(row, col, value, addv, im[i], in[j]);
376: /* PetscCall(MatSetValues_SeqBAIJ(baij->B,1,&row,1,&col,&value,addv)); */
377: }
378: }
379: } else { /* off processor entry */
380: PetscCheck(!mat->nooffprocentries, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Setting off process row %" PetscInt_FMT " even though MatSetOption(,MAT_NO_OFF_PROC_ENTRIES,PETSC_TRUE) was set", im[i]);
381: if (!baij->donotstash) {
382: mat->assembled = PETSC_FALSE;
383: n_loc = 0;
384: for (j = 0; j < n; j++) {
385: if (im[i] / bs > in[j] / bs) continue; /* ignore lower triangular blocks */
386: in_loc[n_loc] = in[j];
387: if (roworiented) {
388: v_loc[n_loc] = v[i * n + j];
389: } else {
390: v_loc[n_loc] = v[j * m + i];
391: }
392: n_loc++;
393: }
394: PetscCall(MatStashValuesRow_Private(&mat->stash, im[i], n_loc, in_loc, v_loc, PETSC_FALSE));
395: }
396: }
397: }
398: PetscFunctionReturn(PETSC_SUCCESS);
399: }
401: static inline PetscErrorCode MatSetValuesBlocked_SeqSBAIJ_Inlined(Mat A, PetscInt row, PetscInt col, const PetscScalar v[], InsertMode is, PetscInt orow, PetscInt ocol)
402: {
403: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)A->data;
404: PetscInt *rp, low, high, t, ii, jj, nrow, i, rmax, N;
405: PetscInt *imax = a->imax, *ai = a->i, *ailen = a->ilen;
406: PetscInt *aj = a->j, nonew = a->nonew, bs2 = a->bs2, bs = A->rmap->bs;
407: PetscBool roworiented = a->roworiented;
408: const PetscScalar *value = v;
409: MatScalar *ap, *aa = a->a, *bap;
411: PetscFunctionBegin;
412: if (col < row) {
413: if (a->ignore_ltriangular) PetscFunctionReturn(PETSC_SUCCESS); /* ignore lower triangular block */
414: else SETERRQ(PETSC_COMM_SELF, PETSC_ERR_USER, "Lower triangular value cannot be set for sbaij format. Ignoring these values, run with -mat_ignore_lower_triangular or call MatSetOption(mat,MAT_IGNORE_LOWER_TRIANGULAR,PETSC_TRUE)");
415: }
416: rp = aj + ai[row];
417: ap = aa + bs2 * ai[row];
418: rmax = imax[row];
419: nrow = ailen[row];
420: value = v;
421: low = 0;
422: high = nrow;
424: while (high - low > 7) {
425: t = (low + high) / 2;
426: if (rp[t] > col) high = t;
427: else low = t;
428: }
429: for (i = low; i < high; i++) {
430: if (rp[i] > col) break;
431: if (rp[i] == col) {
432: bap = ap + bs2 * i;
433: if (roworiented) {
434: if (is == ADD_VALUES) {
435: for (ii = 0; ii < bs; ii++) {
436: for (jj = ii; jj < bs2; jj += bs) bap[jj] += *value++;
437: }
438: } else {
439: for (ii = 0; ii < bs; ii++) {
440: for (jj = ii; jj < bs2; jj += bs) bap[jj] = *value++;
441: }
442: }
443: } else {
444: if (is == ADD_VALUES) {
445: for (ii = 0; ii < bs; ii++) {
446: for (jj = 0; jj < bs; jj++) *bap++ += *value++;
447: }
448: } else {
449: for (ii = 0; ii < bs; ii++) {
450: for (jj = 0; jj < bs; jj++) *bap++ = *value++;
451: }
452: }
453: }
454: goto noinsert2;
455: }
456: }
457: if (nonew == 1) goto noinsert2;
458: PetscCheck(nonew != -1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Inserting a new block index nonzero block (%" PetscInt_FMT ", %" PetscInt_FMT ") in the matrix", orow, ocol);
459: MatSeqXAIJReallocateAIJ(A, a->mbs, bs2, nrow, row, col, rmax, aa, ai, aj, rp, ap, imax, nonew, MatScalar);
460: N = nrow++ - 1;
461: high++;
462: /* shift up all the later entries in this row */
463: PetscCall(PetscArraymove(rp + i + 1, rp + i, N - i + 1));
464: PetscCall(PetscArraymove(ap + bs2 * (i + 1), ap + bs2 * i, bs2 * (N - i + 1)));
465: rp[i] = col;
466: bap = ap + bs2 * i;
467: if (roworiented) {
468: for (ii = 0; ii < bs; ii++) {
469: for (jj = ii; jj < bs2; jj += bs) bap[jj] = *value++;
470: }
471: } else {
472: for (ii = 0; ii < bs; ii++) {
473: for (jj = 0; jj < bs; jj++) *bap++ = *value++;
474: }
475: }
476: noinsert2:;
477: ailen[row] = nrow;
478: PetscFunctionReturn(PETSC_SUCCESS);
479: }
481: /*
482: This routine is exactly duplicated in mpibaij.c
483: */
484: static inline PetscErrorCode MatSetValuesBlocked_SeqBAIJ_Inlined(Mat A, PetscInt row, PetscInt col, const PetscScalar v[], InsertMode is, PetscInt orow, PetscInt ocol)
485: {
486: Mat_SeqBAIJ *a = (Mat_SeqBAIJ *)A->data;
487: PetscInt *rp, low, high, t, ii, jj, nrow, i, rmax, N;
488: PetscInt *imax = a->imax, *ai = a->i, *ailen = a->ilen;
489: PetscInt *aj = a->j, nonew = a->nonew, bs2 = a->bs2, bs = A->rmap->bs;
490: PetscBool roworiented = a->roworiented;
491: const PetscScalar *value = v;
492: MatScalar *ap, *aa = a->a, *bap;
494: PetscFunctionBegin;
495: rp = aj + ai[row];
496: ap = aa + bs2 * ai[row];
497: rmax = imax[row];
498: nrow = ailen[row];
499: low = 0;
500: high = nrow;
501: value = v;
502: while (high - low > 7) {
503: t = (low + high) / 2;
504: if (rp[t] > col) high = t;
505: else low = t;
506: }
507: for (i = low; i < high; i++) {
508: if (rp[i] > col) break;
509: if (rp[i] == col) {
510: bap = ap + bs2 * i;
511: if (roworiented) {
512: if (is == ADD_VALUES) {
513: for (ii = 0; ii < bs; ii++) {
514: for (jj = ii; jj < bs2; jj += bs) bap[jj] += *value++;
515: }
516: } else {
517: for (ii = 0; ii < bs; ii++) {
518: for (jj = ii; jj < bs2; jj += bs) bap[jj] = *value++;
519: }
520: }
521: } else {
522: if (is == ADD_VALUES) {
523: for (ii = 0; ii < bs; ii++, value += bs) {
524: for (jj = 0; jj < bs; jj++) bap[jj] += value[jj];
525: bap += bs;
526: }
527: } else {
528: for (ii = 0; ii < bs; ii++, value += bs) {
529: for (jj = 0; jj < bs; jj++) bap[jj] = value[jj];
530: bap += bs;
531: }
532: }
533: }
534: goto noinsert2;
535: }
536: }
537: if (nonew == 1) goto noinsert2;
538: PetscCheck(nonew != -1, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Inserting a new global block indexed nonzero block (%" PetscInt_FMT ", %" PetscInt_FMT ") in the matrix", orow, ocol);
539: MatSeqXAIJReallocateAIJ(A, a->mbs, bs2, nrow, row, col, rmax, aa, ai, aj, rp, ap, imax, nonew, MatScalar);
540: N = nrow++ - 1;
541: high++;
542: /* shift up all the later entries in this row */
543: PetscCall(PetscArraymove(rp + i + 1, rp + i, N - i + 1));
544: PetscCall(PetscArraymove(ap + bs2 * (i + 1), ap + bs2 * i, bs2 * (N - i + 1)));
545: rp[i] = col;
546: bap = ap + bs2 * i;
547: if (roworiented) {
548: for (ii = 0; ii < bs; ii++) {
549: for (jj = ii; jj < bs2; jj += bs) bap[jj] = *value++;
550: }
551: } else {
552: for (ii = 0; ii < bs; ii++) {
553: for (jj = 0; jj < bs; jj++) *bap++ = *value++;
554: }
555: }
556: noinsert2:;
557: ailen[row] = nrow;
558: PetscFunctionReturn(PETSC_SUCCESS);
559: }
561: /*
562: This routine could be optimized by removing the need for the block copy below and passing stride information
563: to the above inline routines; similarly in MatSetValuesBlocked_MPIBAIJ()
564: */
565: static PetscErrorCode MatSetValuesBlocked_MPISBAIJ(Mat mat, PetscInt m, const PetscInt im[], PetscInt n, const PetscInt in[], const MatScalar v[], InsertMode addv)
566: {
567: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ *)mat->data;
568: const MatScalar *value;
569: MatScalar *barray = baij->barray;
570: PetscBool roworiented = baij->roworiented, ignore_ltriangular = ((Mat_SeqSBAIJ *)baij->A->data)->ignore_ltriangular;
571: PetscInt i, j, ii, jj, row, col, rstart = baij->rstartbs;
572: PetscInt rend = baij->rendbs, cstart = baij->cstartbs, stepval;
573: PetscInt cend = baij->cendbs, bs = mat->rmap->bs, bs2 = baij->bs2;
575: PetscFunctionBegin;
576: if (!barray) {
577: PetscCall(PetscMalloc1(bs2, &barray));
578: baij->barray = barray;
579: }
581: if (roworiented) {
582: stepval = (n - 1) * bs;
583: } else {
584: stepval = (m - 1) * bs;
585: }
586: for (i = 0; i < m; i++) {
587: if (im[i] < 0) continue;
588: PetscCheck(im[i] < baij->Mbs, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Block indexed row too large %" PetscInt_FMT " max %" PetscInt_FMT, im[i], baij->Mbs - 1);
589: if (im[i] >= rstart && im[i] < rend) {
590: row = im[i] - rstart;
591: for (j = 0; j < n; j++) {
592: if (im[i] > in[j]) {
593: if (ignore_ltriangular) continue; /* ignore lower triangular blocks */
594: else SETERRQ(PETSC_COMM_SELF, PETSC_ERR_USER, "Lower triangular value cannot be set for sbaij format. Ignoring these values, run with -mat_ignore_lower_triangular or call MatSetOption(mat,MAT_IGNORE_LOWER_TRIANGULAR,PETSC_TRUE)");
595: }
596: /* If NumCol = 1 then a copy is not required */
597: if ((roworiented) && (n == 1)) {
598: barray = (MatScalar *)v + i * bs2;
599: } else if ((!roworiented) && (m == 1)) {
600: barray = (MatScalar *)v + j * bs2;
601: } else { /* Here a copy is required */
602: if (roworiented) {
603: value = v + i * (stepval + bs) * bs + j * bs;
604: } else {
605: value = v + j * (stepval + bs) * bs + i * bs;
606: }
607: for (ii = 0; ii < bs; ii++, value += stepval) {
608: for (jj = 0; jj < bs; jj++) *barray++ = *value++;
609: }
610: barray -= bs2;
611: }
613: if (in[j] >= cstart && in[j] < cend) {
614: col = in[j] - cstart;
615: PetscCall(MatSetValuesBlocked_SeqSBAIJ_Inlined(baij->A, row, col, barray, addv, im[i], in[j]));
616: } else if (in[j] < 0) {
617: continue;
618: } else {
619: PetscCheck(in[j] < baij->Nbs, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Block indexed column too large %" PetscInt_FMT " max %" PetscInt_FMT, in[j], baij->Nbs - 1);
620: if (mat->was_assembled) {
621: if (!baij->colmap) PetscCall(MatCreateColmap_MPIBAIJ_Private(mat));
623: #if defined(PETSC_USE_CTABLE)
624: PetscCall(PetscHMapIGetWithDefault(baij->colmap, in[j] + 1, 0, &col));
625: col = col < 1 ? -1 : (col - 1) / bs;
626: #else
627: col = baij->colmap[in[j]] < 1 ? -1 : (baij->colmap[in[j]] - 1) / bs;
628: #endif
629: if (col < 0 && !((Mat_SeqBAIJ *)baij->A->data)->nonew) {
630: PetscCall(MatDisAssemble_MPISBAIJ(mat));
631: col = in[j];
632: }
633: } else col = in[j];
634: PetscCall(MatSetValuesBlocked_SeqBAIJ_Inlined(baij->B, row, col, barray, addv, im[i], in[j]));
635: }
636: }
637: } else {
638: PetscCheck(!mat->nooffprocentries, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Setting off process block indexed row %" PetscInt_FMT " even though MatSetOption(,MAT_NO_OFF_PROC_ENTRIES,PETSC_TRUE) was set", im[i]);
639: if (!baij->donotstash) {
640: if (roworiented) {
641: PetscCall(MatStashValuesRowBlocked_Private(&mat->bstash, im[i], n, in, v, m, n, i));
642: } else {
643: PetscCall(MatStashValuesColBlocked_Private(&mat->bstash, im[i], n, in, v, m, n, i));
644: }
645: }
646: }
647: }
648: PetscFunctionReturn(PETSC_SUCCESS);
649: }
651: static PetscErrorCode MatGetValues_MPISBAIJ(Mat mat, PetscInt m, const PetscInt idxm[], PetscInt n, const PetscInt idxn[], PetscScalar v[])
652: {
653: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ *)mat->data;
654: PetscInt bs = mat->rmap->bs, i, j, bsrstart = mat->rmap->rstart, bsrend = mat->rmap->rend;
655: PetscInt bscstart = mat->cmap->rstart, bscend = mat->cmap->rend, row, col, data;
657: PetscFunctionBegin;
658: for (i = 0; i < m; i++) {
659: if (idxm[i] < 0) continue; /* negative row */
660: PetscCheck(idxm[i] < mat->rmap->N, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Row too large: row %" PetscInt_FMT " max %" PetscInt_FMT, idxm[i], mat->rmap->N - 1);
661: if (idxm[i] >= bsrstart && idxm[i] < bsrend) {
662: row = idxm[i] - bsrstart;
663: for (j = 0; j < n; j++) {
664: if (idxn[j] < 0) continue; /* negative column */
665: PetscCheck(idxn[j] < mat->cmap->N, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Column too large: col %" PetscInt_FMT " max %" PetscInt_FMT, idxn[j], mat->cmap->N - 1);
666: if (idxn[j] >= bscstart && idxn[j] < bscend) {
667: col = idxn[j] - bscstart;
668: PetscCall(MatGetValues_SeqSBAIJ(baij->A, 1, &row, 1, &col, v + i * n + j));
669: } else {
670: if (!baij->colmap) PetscCall(MatCreateColmap_MPIBAIJ_Private(mat));
671: #if defined(PETSC_USE_CTABLE)
672: PetscCall(PetscHMapIGetWithDefault(baij->colmap, idxn[j] / bs + 1, 0, &data));
673: data--;
674: #else
675: data = baij->colmap[idxn[j] / bs] - 1;
676: #endif
677: if ((data < 0) || (baij->garray[data / bs] != idxn[j] / bs)) *(v + i * n + j) = 0.0;
678: else {
679: col = data + idxn[j] % bs;
680: PetscCall(MatGetValues_SeqBAIJ(baij->B, 1, &row, 1, &col, v + i * n + j));
681: }
682: }
683: }
684: } else SETERRQ(PETSC_COMM_SELF, PETSC_ERR_SUP, "Only local values currently supported");
685: }
686: PetscFunctionReturn(PETSC_SUCCESS);
687: }
689: static PetscErrorCode MatNorm_MPISBAIJ(Mat mat, NormType type, PetscReal *norm)
690: {
691: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ *)mat->data;
692: PetscReal sum[2], *lnorm2;
694: PetscFunctionBegin;
695: if (baij->size == 1) {
696: PetscCall(MatNorm(baij->A, type, norm));
697: } else {
698: if (type == NORM_FROBENIUS) {
699: PetscCall(PetscMalloc1(2, &lnorm2));
700: PetscCall(MatNorm(baij->A, type, lnorm2));
701: *lnorm2 = (*lnorm2) * (*lnorm2);
702: lnorm2++; /* squar power of norm(A) */
703: PetscCall(MatNorm(baij->B, type, lnorm2));
704: *lnorm2 = (*lnorm2) * (*lnorm2);
705: lnorm2--; /* squar power of norm(B) */
706: PetscCallMPI(MPIU_Allreduce(lnorm2, sum, 2, MPIU_REAL, MPIU_SUM, PetscObjectComm((PetscObject)mat)));
707: *norm = PetscSqrtReal(sum[0] + 2 * sum[1]);
708: PetscCall(PetscFree(lnorm2));
709: } else if (type == NORM_INFINITY || type == NORM_1) { /* max row/column sum */
710: Mat_SeqSBAIJ *amat = (Mat_SeqSBAIJ *)baij->A->data;
711: Mat_SeqBAIJ *bmat = (Mat_SeqBAIJ *)baij->B->data;
712: PetscReal *rsum, *rsum2, vabs;
713: PetscInt *jj, *garray = baij->garray, rstart = baij->rstartbs, nz;
714: PetscInt brow, bcol, col, bs = baij->A->rmap->bs, row, grow, gcol, mbs = amat->mbs;
715: MatScalar *v;
716: PetscMPIInt iN;
718: PetscCall(PetscMalloc2(mat->cmap->N, &rsum, mat->cmap->N, &rsum2));
719: PetscCall(PetscArrayzero(rsum, mat->cmap->N));
720: /* Amat */
721: v = amat->a;
722: jj = amat->j;
723: for (brow = 0; brow < mbs; brow++) {
724: grow = bs * (rstart + brow);
725: nz = amat->i[brow + 1] - amat->i[brow];
726: for (bcol = 0; bcol < nz; bcol++) {
727: gcol = bs * (rstart + *jj);
728: jj++;
729: for (col = 0; col < bs; col++) {
730: for (row = 0; row < bs; row++) {
731: vabs = PetscAbsScalar(*v);
732: v++;
733: rsum[gcol + col] += vabs;
734: /* non-diagonal block */
735: if (bcol > 0 && vabs > 0.0) rsum[grow + row] += vabs;
736: }
737: }
738: }
739: PetscCall(PetscLogFlops(nz * bs * bs));
740: }
741: /* Bmat */
742: v = bmat->a;
743: jj = bmat->j;
744: for (brow = 0; brow < mbs; brow++) {
745: grow = bs * (rstart + brow);
746: nz = bmat->i[brow + 1] - bmat->i[brow];
747: for (bcol = 0; bcol < nz; bcol++) {
748: gcol = bs * garray[*jj];
749: jj++;
750: for (col = 0; col < bs; col++) {
751: for (row = 0; row < bs; row++) {
752: vabs = PetscAbsScalar(*v);
753: v++;
754: rsum[gcol + col] += vabs;
755: rsum[grow + row] += vabs;
756: }
757: }
758: }
759: PetscCall(PetscLogFlops(nz * bs * bs));
760: }
761: PetscCall(PetscMPIIntCast(mat->cmap->N, &iN));
762: PetscCallMPI(MPIU_Allreduce(rsum, rsum2, iN, MPIU_REAL, MPIU_SUM, PetscObjectComm((PetscObject)mat)));
763: *norm = 0.0;
764: for (col = 0; col < mat->cmap->N; col++) {
765: if (rsum2[col] > *norm) *norm = rsum2[col];
766: }
767: PetscCall(PetscFree2(rsum, rsum2));
768: } else SETERRQ(PETSC_COMM_SELF, PETSC_ERR_SUP, "No support for this norm yet");
769: }
770: PetscFunctionReturn(PETSC_SUCCESS);
771: }
773: static PetscErrorCode MatAssemblyBegin_MPISBAIJ(Mat mat, MatAssemblyType mode)
774: {
775: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ *)mat->data;
776: PetscInt nstash, reallocs;
778: PetscFunctionBegin;
779: if (baij->donotstash || mat->nooffprocentries) PetscFunctionReturn(PETSC_SUCCESS);
781: PetscCall(MatStashScatterBegin_Private(mat, &mat->stash, mat->rmap->range));
782: PetscCall(MatStashScatterBegin_Private(mat, &mat->bstash, baij->rangebs));
783: PetscCall(MatStashGetInfo_Private(&mat->stash, &nstash, &reallocs));
784: PetscCall(PetscInfo(mat, "Stash has %" PetscInt_FMT " entries,uses %" PetscInt_FMT " mallocs.\n", nstash, reallocs));
785: PetscCall(MatStashGetInfo_Private(&mat->stash, &nstash, &reallocs));
786: PetscCall(PetscInfo(mat, "Block-Stash has %" PetscInt_FMT " entries, uses %" PetscInt_FMT " mallocs.\n", nstash, reallocs));
787: PetscFunctionReturn(PETSC_SUCCESS);
788: }
790: static PetscErrorCode MatAssemblyEnd_MPISBAIJ(Mat mat, MatAssemblyType mode)
791: {
792: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ *)mat->data;
793: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)baij->A->data;
794: PetscInt i, j, rstart, ncols, flg, bs2 = baij->bs2;
795: PetscInt *row, *col;
796: PetscBool other_disassembled;
797: PetscMPIInt n;
798: PetscBool r1, r2, r3;
799: MatScalar *val;
801: /* do not use 'b=(Mat_SeqBAIJ*)baij->B->data' as B can be reset in disassembly */
802: PetscFunctionBegin;
803: if (!baij->donotstash && !mat->nooffprocentries) {
804: while (1) {
805: PetscCall(MatStashScatterGetMesg_Private(&mat->stash, &n, &row, &col, &val, &flg));
806: if (!flg) break;
808: for (i = 0; i < n;) {
809: /* Now identify the consecutive vals belonging to the same row */
810: for (j = i, rstart = row[j]; j < n; j++) {
811: if (row[j] != rstart) break;
812: }
813: if (j < n) ncols = j - i;
814: else ncols = n - i;
815: /* Now assemble all these values with a single function call */
816: PetscCall(MatSetValues_MPISBAIJ(mat, 1, row + i, ncols, col + i, val + i, mat->insertmode));
817: i = j;
818: }
819: }
820: PetscCall(MatStashScatterEnd_Private(&mat->stash));
821: /* Now process the block-stash. Since the values are stashed column-oriented,
822: set the row-oriented flag to column-oriented, and after MatSetValues()
823: restore the original flags */
824: r1 = baij->roworiented;
825: r2 = a->roworiented;
826: r3 = ((Mat_SeqBAIJ *)baij->B->data)->roworiented;
828: baij->roworiented = PETSC_FALSE;
829: a->roworiented = PETSC_FALSE;
831: ((Mat_SeqBAIJ *)baij->B->data)->roworiented = PETSC_FALSE; /* b->roworinted */
832: while (1) {
833: PetscCall(MatStashScatterGetMesg_Private(&mat->bstash, &n, &row, &col, &val, &flg));
834: if (!flg) break;
836: for (i = 0; i < n;) {
837: /* Now identify the consecutive vals belonging to the same row */
838: for (j = i, rstart = row[j]; j < n; j++) {
839: if (row[j] != rstart) break;
840: }
841: if (j < n) ncols = j - i;
842: else ncols = n - i;
843: PetscCall(MatSetValuesBlocked_MPISBAIJ(mat, 1, row + i, ncols, col + i, val + i * bs2, mat->insertmode));
844: i = j;
845: }
846: }
847: PetscCall(MatStashScatterEnd_Private(&mat->bstash));
849: baij->roworiented = r1;
850: a->roworiented = r2;
852: ((Mat_SeqBAIJ *)baij->B->data)->roworiented = r3; /* b->roworinted */
853: }
855: PetscCall(MatAssemblyBegin(baij->A, mode));
856: PetscCall(MatAssemblyEnd(baij->A, mode));
858: /* determine if any processor has disassembled, if so we must
859: also disassemble ourselves, in order that we may reassemble. */
860: /*
861: if nonzero structure of submatrix B cannot change then we know that
862: no processor disassembled thus we can skip this stuff
863: */
864: if (!((Mat_SeqBAIJ *)baij->B->data)->nonew) {
865: PetscCallMPI(MPIU_Allreduce(&mat->was_assembled, &other_disassembled, 1, MPIU_BOOL, MPI_LAND, PetscObjectComm((PetscObject)mat)));
866: if (mat->was_assembled && !other_disassembled) PetscCall(MatDisAssemble_MPISBAIJ(mat));
867: }
869: if (!mat->was_assembled && mode == MAT_FINAL_ASSEMBLY) { PetscCall(MatSetUpMultiply_MPISBAIJ(mat)); /* setup Mvctx and sMvctx */ }
870: PetscCall(MatAssemblyBegin(baij->B, mode));
871: PetscCall(MatAssemblyEnd(baij->B, mode));
873: PetscCall(PetscFree2(baij->rowvalues, baij->rowindices));
875: baij->rowvalues = NULL;
877: /* if no new nonzero locations are allowed in matrix then only set the matrix state the first time through */
878: if ((!mat->was_assembled && mode == MAT_FINAL_ASSEMBLY) || !((Mat_SeqBAIJ *)baij->A->data)->nonew) {
879: PetscObjectState state = baij->A->nonzerostate + baij->B->nonzerostate;
880: PetscCallMPI(MPIU_Allreduce(&state, &mat->nonzerostate, 1, MPIU_INT64, MPI_SUM, PetscObjectComm((PetscObject)mat)));
881: }
882: PetscFunctionReturn(PETSC_SUCCESS);
883: }
885: extern PetscErrorCode MatSetValues_MPIBAIJ(Mat, PetscInt, const PetscInt[], PetscInt, const PetscInt[], const PetscScalar[], InsertMode);
886: #include <petscdraw.h>
887: static PetscErrorCode MatView_MPISBAIJ_ASCIIorDraworSocket(Mat mat, PetscViewer viewer)
888: {
889: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ *)mat->data;
890: PetscInt bs = mat->rmap->bs;
891: PetscMPIInt rank = baij->rank;
892: PetscBool iascii, isdraw;
893: PetscViewer sviewer;
894: PetscViewerFormat format;
896: PetscFunctionBegin;
897: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &iascii));
898: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERDRAW, &isdraw));
899: if (iascii) {
900: PetscCall(PetscViewerGetFormat(viewer, &format));
901: if (format == PETSC_VIEWER_ASCII_INFO_DETAIL) {
902: MatInfo info;
903: PetscCallMPI(MPI_Comm_rank(PetscObjectComm((PetscObject)mat), &rank));
904: PetscCall(MatGetInfo(mat, MAT_LOCAL, &info));
905: PetscCall(PetscViewerASCIIPushSynchronized(viewer));
906: PetscCall(PetscViewerASCIISynchronizedPrintf(viewer, "[%d] Local rows %" PetscInt_FMT " nz %" PetscInt_FMT " nz alloced %" PetscInt_FMT " bs %" PetscInt_FMT " mem %g\n", rank, mat->rmap->n, (PetscInt)info.nz_used, (PetscInt)info.nz_allocated,
907: mat->rmap->bs, info.memory));
908: PetscCall(MatGetInfo(baij->A, MAT_LOCAL, &info));
909: PetscCall(PetscViewerASCIISynchronizedPrintf(viewer, "[%d] on-diagonal part: nz %" PetscInt_FMT " \n", rank, (PetscInt)info.nz_used));
910: PetscCall(MatGetInfo(baij->B, MAT_LOCAL, &info));
911: PetscCall(PetscViewerASCIISynchronizedPrintf(viewer, "[%d] off-diagonal part: nz %" PetscInt_FMT " \n", rank, (PetscInt)info.nz_used));
912: PetscCall(PetscViewerFlush(viewer));
913: PetscCall(PetscViewerASCIIPopSynchronized(viewer));
914: PetscCall(PetscViewerASCIIPrintf(viewer, "Information on VecScatter used in matrix-vector product: \n"));
915: PetscCall(VecScatterView(baij->Mvctx, viewer));
916: PetscFunctionReturn(PETSC_SUCCESS);
917: } else if (format == PETSC_VIEWER_ASCII_INFO) {
918: PetscCall(PetscViewerASCIIPrintf(viewer, " block size is %" PetscInt_FMT "\n", bs));
919: PetscFunctionReturn(PETSC_SUCCESS);
920: } else if (format == PETSC_VIEWER_ASCII_FACTOR_INFO) {
921: PetscFunctionReturn(PETSC_SUCCESS);
922: }
923: }
925: if (isdraw) {
926: PetscDraw draw;
927: PetscBool isnull;
928: PetscCall(PetscViewerDrawGetDraw(viewer, 0, &draw));
929: PetscCall(PetscDrawIsNull(draw, &isnull));
930: if (isnull) PetscFunctionReturn(PETSC_SUCCESS);
931: }
933: {
934: /* assemble the entire matrix onto first processor. */
935: Mat A;
936: Mat_SeqSBAIJ *Aloc;
937: Mat_SeqBAIJ *Bloc;
938: PetscInt M = mat->rmap->N, N = mat->cmap->N, *ai, *aj, col, i, j, k, *rvals, mbs = baij->mbs;
939: MatScalar *a;
940: const char *matname;
942: /* Should this be the same type as mat? */
943: PetscCall(MatCreate(PetscObjectComm((PetscObject)mat), &A));
944: if (rank == 0) {
945: PetscCall(MatSetSizes(A, M, N, M, N));
946: } else {
947: PetscCall(MatSetSizes(A, 0, 0, M, N));
948: }
949: PetscCall(MatSetType(A, MATMPISBAIJ));
950: PetscCall(MatMPISBAIJSetPreallocation(A, mat->rmap->bs, 0, NULL, 0, NULL));
951: PetscCall(MatSetOption(A, MAT_NEW_NONZERO_LOCATION_ERR, PETSC_FALSE));
953: /* copy over the A part */
954: Aloc = (Mat_SeqSBAIJ *)baij->A->data;
955: ai = Aloc->i;
956: aj = Aloc->j;
957: a = Aloc->a;
958: PetscCall(PetscMalloc1(bs, &rvals));
960: for (i = 0; i < mbs; i++) {
961: rvals[0] = bs * (baij->rstartbs + i);
962: for (j = 1; j < bs; j++) rvals[j] = rvals[j - 1] + 1;
963: for (j = ai[i]; j < ai[i + 1]; j++) {
964: col = (baij->cstartbs + aj[j]) * bs;
965: for (k = 0; k < bs; k++) {
966: PetscCall(MatSetValues_MPISBAIJ(A, bs, rvals, 1, &col, a, INSERT_VALUES));
967: col++;
968: a += bs;
969: }
970: }
971: }
972: /* copy over the B part */
973: Bloc = (Mat_SeqBAIJ *)baij->B->data;
974: ai = Bloc->i;
975: aj = Bloc->j;
976: a = Bloc->a;
977: for (i = 0; i < mbs; i++) {
978: rvals[0] = bs * (baij->rstartbs + i);
979: for (j = 1; j < bs; j++) rvals[j] = rvals[j - 1] + 1;
980: for (j = ai[i]; j < ai[i + 1]; j++) {
981: col = baij->garray[aj[j]] * bs;
982: for (k = 0; k < bs; k++) {
983: PetscCall(MatSetValues_MPIBAIJ(A, bs, rvals, 1, &col, a, INSERT_VALUES));
984: col++;
985: a += bs;
986: }
987: }
988: }
989: PetscCall(PetscFree(rvals));
990: PetscCall(MatAssemblyBegin(A, MAT_FINAL_ASSEMBLY));
991: PetscCall(MatAssemblyEnd(A, MAT_FINAL_ASSEMBLY));
992: /*
993: Everyone has to call to draw the matrix since the graphics waits are
994: synchronized across all processors that share the PetscDraw object
995: */
996: PetscCall(PetscViewerGetSubViewer(viewer, PETSC_COMM_SELF, &sviewer));
997: if (((PetscObject)mat)->name) PetscCall(PetscObjectGetName((PetscObject)mat, &matname));
998: if (rank == 0) {
999: if (((PetscObject)mat)->name) PetscCall(PetscObjectSetName((PetscObject)((Mat_MPISBAIJ *)A->data)->A, matname));
1000: PetscCall(MatView_SeqSBAIJ(((Mat_MPISBAIJ *)A->data)->A, sviewer));
1001: }
1002: PetscCall(PetscViewerRestoreSubViewer(viewer, PETSC_COMM_SELF, &sviewer));
1003: PetscCall(MatDestroy(&A));
1004: }
1005: PetscFunctionReturn(PETSC_SUCCESS);
1006: }
1008: /* Used for both MPIBAIJ and MPISBAIJ matrices */
1009: #define MatView_MPISBAIJ_Binary MatView_MPIBAIJ_Binary
1011: static PetscErrorCode MatView_MPISBAIJ(Mat mat, PetscViewer viewer)
1012: {
1013: PetscBool iascii, isdraw, issocket, isbinary;
1015: PetscFunctionBegin;
1016: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERASCII, &iascii));
1017: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERDRAW, &isdraw));
1018: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERSOCKET, &issocket));
1019: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERBINARY, &isbinary));
1020: if (iascii || isdraw || issocket) {
1021: PetscCall(MatView_MPISBAIJ_ASCIIorDraworSocket(mat, viewer));
1022: } else if (isbinary) PetscCall(MatView_MPISBAIJ_Binary(mat, viewer));
1023: PetscFunctionReturn(PETSC_SUCCESS);
1024: }
1026: #if defined(PETSC_USE_COMPLEX)
1027: static PetscErrorCode MatMult_MPISBAIJ_Hermitian(Mat A, Vec xx, Vec yy)
1028: {
1029: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1030: PetscInt mbs = a->mbs, bs = A->rmap->bs;
1031: PetscScalar *from;
1032: const PetscScalar *x;
1034: PetscFunctionBegin;
1035: /* diagonal part */
1036: PetscCall((*a->A->ops->mult)(a->A, xx, a->slvec1a));
1037: /* since a->slvec1b shares memory (dangerously) with a->slec1 changes to a->slec1 will affect it */
1038: PetscCall(PetscObjectStateIncrease((PetscObject)a->slvec1b));
1039: PetscCall(VecZeroEntries(a->slvec1b));
1041: /* subdiagonal part */
1042: PetscCheck(a->B->ops->multhermitiantranspose, PetscObjectComm((PetscObject)a->B), PETSC_ERR_SUP, "Not for type %s", ((PetscObject)a->B)->type_name);
1043: PetscCall((*a->B->ops->multhermitiantranspose)(a->B, xx, a->slvec0b));
1045: /* copy x into the vec slvec0 */
1046: PetscCall(VecGetArray(a->slvec0, &from));
1047: PetscCall(VecGetArrayRead(xx, &x));
1049: PetscCall(PetscArraycpy(from, x, bs * mbs));
1050: PetscCall(VecRestoreArray(a->slvec0, &from));
1051: PetscCall(VecRestoreArrayRead(xx, &x));
1053: PetscCall(VecScatterBegin(a->sMvctx, a->slvec0, a->slvec1, ADD_VALUES, SCATTER_FORWARD));
1054: PetscCall(VecScatterEnd(a->sMvctx, a->slvec0, a->slvec1, ADD_VALUES, SCATTER_FORWARD));
1055: /* supperdiagonal part */
1056: PetscCall((*a->B->ops->multadd)(a->B, a->slvec1b, a->slvec1a, yy));
1057: PetscFunctionReturn(PETSC_SUCCESS);
1058: }
1059: #endif
1061: static PetscErrorCode MatMult_MPISBAIJ(Mat A, Vec xx, Vec yy)
1062: {
1063: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1064: PetscInt mbs = a->mbs, bs = A->rmap->bs;
1065: PetscScalar *from;
1066: const PetscScalar *x;
1068: PetscFunctionBegin;
1069: /* diagonal part */
1070: PetscCall((*a->A->ops->mult)(a->A, xx, a->slvec1a));
1071: /* since a->slvec1b shares memory (dangerously) with a->slec1 changes to a->slec1 will affect it */
1072: PetscCall(PetscObjectStateIncrease((PetscObject)a->slvec1b));
1073: PetscCall(VecZeroEntries(a->slvec1b));
1075: /* subdiagonal part */
1076: PetscCall((*a->B->ops->multtranspose)(a->B, xx, a->slvec0b));
1078: /* copy x into the vec slvec0 */
1079: PetscCall(VecGetArray(a->slvec0, &from));
1080: PetscCall(VecGetArrayRead(xx, &x));
1082: PetscCall(PetscArraycpy(from, x, bs * mbs));
1083: PetscCall(VecRestoreArray(a->slvec0, &from));
1084: PetscCall(VecRestoreArrayRead(xx, &x));
1086: PetscCall(VecScatterBegin(a->sMvctx, a->slvec0, a->slvec1, ADD_VALUES, SCATTER_FORWARD));
1087: PetscCall(VecScatterEnd(a->sMvctx, a->slvec0, a->slvec1, ADD_VALUES, SCATTER_FORWARD));
1088: /* supperdiagonal part */
1089: PetscCall((*a->B->ops->multadd)(a->B, a->slvec1b, a->slvec1a, yy));
1090: PetscFunctionReturn(PETSC_SUCCESS);
1091: }
1093: #if PetscDefined(USE_COMPLEX)
1094: static PetscErrorCode MatMultAdd_MPISBAIJ_Hermitian(Mat A, Vec xx, Vec yy, Vec zz)
1095: {
1096: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1097: PetscInt mbs = a->mbs, bs = A->rmap->bs;
1098: PetscScalar *from;
1099: const PetscScalar *x;
1101: PetscFunctionBegin;
1102: /* diagonal part */
1103: PetscCall((*a->A->ops->multadd)(a->A, xx, yy, a->slvec1a));
1104: PetscCall(PetscObjectStateIncrease((PetscObject)a->slvec1b));
1105: PetscCall(VecZeroEntries(a->slvec1b));
1107: /* subdiagonal part */
1108: PetscCheck(a->B->ops->multhermitiantranspose, PetscObjectComm((PetscObject)a->B), PETSC_ERR_SUP, "Not for type %s", ((PetscObject)a->B)->type_name);
1109: PetscCall((*a->B->ops->multhermitiantranspose)(a->B, xx, a->slvec0b));
1111: /* copy x into the vec slvec0 */
1112: PetscCall(VecGetArray(a->slvec0, &from));
1113: PetscCall(VecGetArrayRead(xx, &x));
1114: PetscCall(PetscArraycpy(from, x, bs * mbs));
1115: PetscCall(VecRestoreArray(a->slvec0, &from));
1117: PetscCall(VecScatterBegin(a->sMvctx, a->slvec0, a->slvec1, ADD_VALUES, SCATTER_FORWARD));
1118: PetscCall(VecRestoreArrayRead(xx, &x));
1119: PetscCall(VecScatterEnd(a->sMvctx, a->slvec0, a->slvec1, ADD_VALUES, SCATTER_FORWARD));
1121: /* supperdiagonal part */
1122: PetscCall((*a->B->ops->multadd)(a->B, a->slvec1b, a->slvec1a, zz));
1123: PetscFunctionReturn(PETSC_SUCCESS);
1124: }
1125: #endif
1127: static PetscErrorCode MatMultAdd_MPISBAIJ(Mat A, Vec xx, Vec yy, Vec zz)
1128: {
1129: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1130: PetscInt mbs = a->mbs, bs = A->rmap->bs;
1131: PetscScalar *from;
1132: const PetscScalar *x;
1134: PetscFunctionBegin;
1135: /* diagonal part */
1136: PetscCall((*a->A->ops->multadd)(a->A, xx, yy, a->slvec1a));
1137: PetscCall(PetscObjectStateIncrease((PetscObject)a->slvec1b));
1138: PetscCall(VecZeroEntries(a->slvec1b));
1140: /* subdiagonal part */
1141: PetscCall((*a->B->ops->multtranspose)(a->B, xx, a->slvec0b));
1143: /* copy x into the vec slvec0 */
1144: PetscCall(VecGetArray(a->slvec0, &from));
1145: PetscCall(VecGetArrayRead(xx, &x));
1146: PetscCall(PetscArraycpy(from, x, bs * mbs));
1147: PetscCall(VecRestoreArray(a->slvec0, &from));
1149: PetscCall(VecScatterBegin(a->sMvctx, a->slvec0, a->slvec1, ADD_VALUES, SCATTER_FORWARD));
1150: PetscCall(VecRestoreArrayRead(xx, &x));
1151: PetscCall(VecScatterEnd(a->sMvctx, a->slvec0, a->slvec1, ADD_VALUES, SCATTER_FORWARD));
1153: /* supperdiagonal part */
1154: PetscCall((*a->B->ops->multadd)(a->B, a->slvec1b, a->slvec1a, zz));
1155: PetscFunctionReturn(PETSC_SUCCESS);
1156: }
1158: /*
1159: This only works correctly for square matrices where the subblock A->A is the
1160: diagonal block
1161: */
1162: static PetscErrorCode MatGetDiagonal_MPISBAIJ(Mat A, Vec v)
1163: {
1164: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1166: PetscFunctionBegin;
1167: /* PetscCheck(a->rmap->N == a->cmap->N,PETSC_COMM_SELF,PETSC_ERR_SUP,"Supports only square matrix where A->A is diag block"); */
1168: PetscCall(MatGetDiagonal(a->A, v));
1169: PetscFunctionReturn(PETSC_SUCCESS);
1170: }
1172: static PetscErrorCode MatScale_MPISBAIJ(Mat A, PetscScalar aa)
1173: {
1174: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1176: PetscFunctionBegin;
1177: PetscCall(MatScale(a->A, aa));
1178: PetscCall(MatScale(a->B, aa));
1179: PetscFunctionReturn(PETSC_SUCCESS);
1180: }
1182: static PetscErrorCode MatGetRow_MPISBAIJ(Mat matin, PetscInt row, PetscInt *nz, PetscInt **idx, PetscScalar **v)
1183: {
1184: Mat_MPISBAIJ *mat = (Mat_MPISBAIJ *)matin->data;
1185: PetscScalar *vworkA, *vworkB, **pvA, **pvB, *v_p;
1186: PetscInt bs = matin->rmap->bs, bs2 = mat->bs2, i, *cworkA, *cworkB, **pcA, **pcB;
1187: PetscInt nztot, nzA, nzB, lrow, brstart = matin->rmap->rstart, brend = matin->rmap->rend;
1188: PetscInt *cmap, *idx_p, cstart = mat->rstartbs;
1190: PetscFunctionBegin;
1191: PetscCheck(!mat->getrowactive, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Already active");
1192: mat->getrowactive = PETSC_TRUE;
1194: if (!mat->rowvalues && (idx || v)) {
1195: /*
1196: allocate enough space to hold information from the longest row.
1197: */
1198: Mat_SeqSBAIJ *Aa = (Mat_SeqSBAIJ *)mat->A->data;
1199: Mat_SeqBAIJ *Ba = (Mat_SeqBAIJ *)mat->B->data;
1200: PetscInt max = 1, mbs = mat->mbs, tmp;
1201: for (i = 0; i < mbs; i++) {
1202: tmp = Aa->i[i + 1] - Aa->i[i] + Ba->i[i + 1] - Ba->i[i]; /* row length */
1203: if (max < tmp) max = tmp;
1204: }
1205: PetscCall(PetscMalloc2(max * bs2, &mat->rowvalues, max * bs2, &mat->rowindices));
1206: }
1208: PetscCheck(row >= brstart && row < brend, PETSC_COMM_SELF, PETSC_ERR_SUP, "Only local rows");
1209: lrow = row - brstart; /* local row index */
1211: pvA = &vworkA;
1212: pcA = &cworkA;
1213: pvB = &vworkB;
1214: pcB = &cworkB;
1215: if (!v) {
1216: pvA = NULL;
1217: pvB = NULL;
1218: }
1219: if (!idx) {
1220: pcA = NULL;
1221: if (!v) pcB = NULL;
1222: }
1223: PetscCall((*mat->A->ops->getrow)(mat->A, lrow, &nzA, pcA, pvA));
1224: PetscCall((*mat->B->ops->getrow)(mat->B, lrow, &nzB, pcB, pvB));
1225: nztot = nzA + nzB;
1227: cmap = mat->garray;
1228: if (v || idx) {
1229: if (nztot) {
1230: /* Sort by increasing column numbers, assuming A and B already sorted */
1231: PetscInt imark = -1;
1232: if (v) {
1233: *v = v_p = mat->rowvalues;
1234: for (i = 0; i < nzB; i++) {
1235: if (cmap[cworkB[i] / bs] < cstart) v_p[i] = vworkB[i];
1236: else break;
1237: }
1238: imark = i;
1239: for (i = 0; i < nzA; i++) v_p[imark + i] = vworkA[i];
1240: for (i = imark; i < nzB; i++) v_p[nzA + i] = vworkB[i];
1241: }
1242: if (idx) {
1243: *idx = idx_p = mat->rowindices;
1244: if (imark > -1) {
1245: for (i = 0; i < imark; i++) idx_p[i] = cmap[cworkB[i] / bs] * bs + cworkB[i] % bs;
1246: } else {
1247: for (i = 0; i < nzB; i++) {
1248: if (cmap[cworkB[i] / bs] < cstart) idx_p[i] = cmap[cworkB[i] / bs] * bs + cworkB[i] % bs;
1249: else break;
1250: }
1251: imark = i;
1252: }
1253: for (i = 0; i < nzA; i++) idx_p[imark + i] = cstart * bs + cworkA[i];
1254: for (i = imark; i < nzB; i++) idx_p[nzA + i] = cmap[cworkB[i] / bs] * bs + cworkB[i] % bs;
1255: }
1256: } else {
1257: if (idx) *idx = NULL;
1258: if (v) *v = NULL;
1259: }
1260: }
1261: *nz = nztot;
1262: PetscCall((*mat->A->ops->restorerow)(mat->A, lrow, &nzA, pcA, pvA));
1263: PetscCall((*mat->B->ops->restorerow)(mat->B, lrow, &nzB, pcB, pvB));
1264: PetscFunctionReturn(PETSC_SUCCESS);
1265: }
1267: static PetscErrorCode MatRestoreRow_MPISBAIJ(Mat mat, PetscInt row, PetscInt *nz, PetscInt **idx, PetscScalar **v)
1268: {
1269: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ *)mat->data;
1271: PetscFunctionBegin;
1272: PetscCheck(baij->getrowactive, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "MatGetRow() must be called first");
1273: baij->getrowactive = PETSC_FALSE;
1274: PetscFunctionReturn(PETSC_SUCCESS);
1275: }
1277: static PetscErrorCode MatGetRowUpperTriangular_MPISBAIJ(Mat A)
1278: {
1279: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1280: Mat_SeqSBAIJ *aA = (Mat_SeqSBAIJ *)a->A->data;
1282: PetscFunctionBegin;
1283: aA->getrow_utriangular = PETSC_TRUE;
1284: PetscFunctionReturn(PETSC_SUCCESS);
1285: }
1286: static PetscErrorCode MatRestoreRowUpperTriangular_MPISBAIJ(Mat A)
1287: {
1288: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1289: Mat_SeqSBAIJ *aA = (Mat_SeqSBAIJ *)a->A->data;
1291: PetscFunctionBegin;
1292: aA->getrow_utriangular = PETSC_FALSE;
1293: PetscFunctionReturn(PETSC_SUCCESS);
1294: }
1296: static PetscErrorCode MatConjugate_MPISBAIJ(Mat mat)
1297: {
1298: PetscFunctionBegin;
1299: if (PetscDefined(USE_COMPLEX)) {
1300: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)mat->data;
1302: PetscCall(MatConjugate(a->A));
1303: PetscCall(MatConjugate(a->B));
1304: }
1305: PetscFunctionReturn(PETSC_SUCCESS);
1306: }
1308: static PetscErrorCode MatRealPart_MPISBAIJ(Mat A)
1309: {
1310: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1312: PetscFunctionBegin;
1313: PetscCall(MatRealPart(a->A));
1314: PetscCall(MatRealPart(a->B));
1315: PetscFunctionReturn(PETSC_SUCCESS);
1316: }
1318: static PetscErrorCode MatImaginaryPart_MPISBAIJ(Mat A)
1319: {
1320: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1322: PetscFunctionBegin;
1323: PetscCall(MatImaginaryPart(a->A));
1324: PetscCall(MatImaginaryPart(a->B));
1325: PetscFunctionReturn(PETSC_SUCCESS);
1326: }
1328: /* Check if isrow is a subset of iscol_local, called by MatCreateSubMatrix_MPISBAIJ()
1329: Input: isrow - distributed(parallel),
1330: iscol_local - locally owned (seq)
1331: */
1332: static PetscErrorCode ISEqual_private(IS isrow, IS iscol_local, PetscBool *flg)
1333: {
1334: PetscInt sz1, sz2, *a1, *a2, i, j, k, nmatch;
1335: const PetscInt *ptr1, *ptr2;
1337: PetscFunctionBegin;
1338: *flg = PETSC_FALSE;
1339: PetscCall(ISGetLocalSize(isrow, &sz1));
1340: PetscCall(ISGetLocalSize(iscol_local, &sz2));
1341: if (sz1 > sz2) PetscFunctionReturn(PETSC_SUCCESS);
1343: PetscCall(ISGetIndices(isrow, &ptr1));
1344: PetscCall(ISGetIndices(iscol_local, &ptr2));
1346: PetscCall(PetscMalloc1(sz1, &a1));
1347: PetscCall(PetscMalloc1(sz2, &a2));
1348: PetscCall(PetscArraycpy(a1, ptr1, sz1));
1349: PetscCall(PetscArraycpy(a2, ptr2, sz2));
1350: PetscCall(PetscSortInt(sz1, a1));
1351: PetscCall(PetscSortInt(sz2, a2));
1353: nmatch = 0;
1354: k = 0;
1355: for (i = 0; i < sz1; i++) {
1356: for (j = k; j < sz2; j++) {
1357: if (a1[i] == a2[j]) {
1358: k = j;
1359: nmatch++;
1360: break;
1361: }
1362: }
1363: }
1364: PetscCall(ISRestoreIndices(isrow, &ptr1));
1365: PetscCall(ISRestoreIndices(iscol_local, &ptr2));
1366: PetscCall(PetscFree(a1));
1367: PetscCall(PetscFree(a2));
1368: if (nmatch < sz1) {
1369: *flg = PETSC_FALSE;
1370: } else {
1371: *flg = PETSC_TRUE;
1372: }
1373: PetscFunctionReturn(PETSC_SUCCESS);
1374: }
1376: static PetscErrorCode MatCreateSubMatrix_MPISBAIJ(Mat mat, IS isrow, IS iscol, MatReuse call, Mat *newmat)
1377: {
1378: Mat C[2];
1379: IS iscol_local, isrow_local;
1380: PetscInt csize, csize_local, rsize;
1381: PetscBool isequal, issorted, isidentity = PETSC_FALSE;
1383: PetscFunctionBegin;
1384: PetscCall(ISGetLocalSize(iscol, &csize));
1385: PetscCall(ISGetLocalSize(isrow, &rsize));
1386: if (call == MAT_REUSE_MATRIX) {
1387: PetscCall(PetscObjectQuery((PetscObject)*newmat, "ISAllGather", (PetscObject *)&iscol_local));
1388: PetscCheck(iscol_local, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Submatrix passed in was not used before, cannot reuse");
1389: } else {
1390: PetscCall(ISAllGather(iscol, &iscol_local));
1391: PetscCall(ISSorted(iscol_local, &issorted));
1392: PetscCheck(issorted, PETSC_COMM_SELF, PETSC_ERR_ARG_INCOMP, "For symmetric format, iscol must be sorted");
1393: }
1394: PetscCall(ISEqual_private(isrow, iscol_local, &isequal));
1395: if (!isequal) {
1396: PetscCall(ISGetLocalSize(iscol_local, &csize_local));
1397: isidentity = (PetscBool)(mat->cmap->N == csize_local);
1398: if (!isidentity) {
1399: if (call == MAT_REUSE_MATRIX) {
1400: PetscCall(PetscObjectQuery((PetscObject)*newmat, "ISAllGather_other", (PetscObject *)&isrow_local));
1401: PetscCheck(isrow_local, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONGSTATE, "Submatrix passed in was not used before, cannot reuse");
1402: } else {
1403: PetscCall(ISAllGather(isrow, &isrow_local));
1404: PetscCall(ISSorted(isrow_local, &issorted));
1405: PetscCheck(issorted, PETSC_COMM_SELF, PETSC_ERR_ARG_INCOMP, "For symmetric format, isrow must be sorted");
1406: }
1407: }
1408: }
1409: /* now call MatCreateSubMatrix_MPIBAIJ() */
1410: PetscCall(MatCreateSubMatrix_MPIBAIJ_Private(mat, isrow, iscol_local, csize, isequal || isidentity ? call : MAT_INITIAL_MATRIX, isequal || isidentity ? newmat : C, (PetscBool)(isequal || isidentity)));
1411: if (!isequal && !isidentity) {
1412: if (call == MAT_INITIAL_MATRIX) {
1413: IS intersect;
1414: PetscInt ni;
1416: PetscCall(ISIntersect(isrow_local, iscol_local, &intersect));
1417: PetscCall(ISGetLocalSize(intersect, &ni));
1418: PetscCall(ISDestroy(&intersect));
1419: PetscCheck(ni == 0, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Cannot create such a submatrix: for symmetric format, when requesting an off-diagonal submatrix, isrow and iscol should have an empty intersection (number of common indices is %" PetscInt_FMT ")", ni);
1420: }
1421: PetscCall(MatCreateSubMatrix_MPIBAIJ_Private(mat, iscol, isrow_local, rsize, MAT_INITIAL_MATRIX, C + 1, PETSC_FALSE));
1422: PetscCall(MatTranspose(C[1], MAT_INPLACE_MATRIX, C + 1));
1423: PetscCall(MatAXPY(C[0], 1.0, C[1], DIFFERENT_NONZERO_PATTERN));
1424: if (call == MAT_REUSE_MATRIX) PetscCall(MatCopy(C[0], *newmat, SAME_NONZERO_PATTERN));
1425: else if (mat->rmap->bs == 1) PetscCall(MatConvert(C[0], MATAIJ, MAT_INITIAL_MATRIX, newmat));
1426: else PetscCall(MatCopy(C[0], *newmat, SAME_NONZERO_PATTERN));
1427: PetscCall(MatDestroy(C));
1428: PetscCall(MatDestroy(C + 1));
1429: }
1430: if (call == MAT_INITIAL_MATRIX) {
1431: if (!isequal && !isidentity) {
1432: PetscCall(PetscObjectCompose((PetscObject)*newmat, "ISAllGather_other", (PetscObject)isrow_local));
1433: PetscCall(ISDestroy(&isrow_local));
1434: }
1435: PetscCall(PetscObjectCompose((PetscObject)*newmat, "ISAllGather", (PetscObject)iscol_local));
1436: PetscCall(ISDestroy(&iscol_local));
1437: }
1438: PetscFunctionReturn(PETSC_SUCCESS);
1439: }
1441: static PetscErrorCode MatZeroEntries_MPISBAIJ(Mat A)
1442: {
1443: Mat_MPISBAIJ *l = (Mat_MPISBAIJ *)A->data;
1445: PetscFunctionBegin;
1446: PetscCall(MatZeroEntries(l->A));
1447: PetscCall(MatZeroEntries(l->B));
1448: PetscFunctionReturn(PETSC_SUCCESS);
1449: }
1451: static PetscErrorCode MatGetInfo_MPISBAIJ(Mat matin, MatInfoType flag, MatInfo *info)
1452: {
1453: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)matin->data;
1454: Mat A = a->A, B = a->B;
1455: PetscLogDouble isend[5], irecv[5];
1457: PetscFunctionBegin;
1458: info->block_size = (PetscReal)matin->rmap->bs;
1460: PetscCall(MatGetInfo(A, MAT_LOCAL, info));
1462: isend[0] = info->nz_used;
1463: isend[1] = info->nz_allocated;
1464: isend[2] = info->nz_unneeded;
1465: isend[3] = info->memory;
1466: isend[4] = info->mallocs;
1468: PetscCall(MatGetInfo(B, MAT_LOCAL, info));
1470: isend[0] += info->nz_used;
1471: isend[1] += info->nz_allocated;
1472: isend[2] += info->nz_unneeded;
1473: isend[3] += info->memory;
1474: isend[4] += info->mallocs;
1475: if (flag == MAT_LOCAL) {
1476: info->nz_used = isend[0];
1477: info->nz_allocated = isend[1];
1478: info->nz_unneeded = isend[2];
1479: info->memory = isend[3];
1480: info->mallocs = isend[4];
1481: } else if (flag == MAT_GLOBAL_MAX) {
1482: PetscCallMPI(MPIU_Allreduce(isend, irecv, 5, MPIU_PETSCLOGDOUBLE, MPI_MAX, PetscObjectComm((PetscObject)matin)));
1484: info->nz_used = irecv[0];
1485: info->nz_allocated = irecv[1];
1486: info->nz_unneeded = irecv[2];
1487: info->memory = irecv[3];
1488: info->mallocs = irecv[4];
1489: } else if (flag == MAT_GLOBAL_SUM) {
1490: PetscCallMPI(MPIU_Allreduce(isend, irecv, 5, MPIU_PETSCLOGDOUBLE, MPI_SUM, PetscObjectComm((PetscObject)matin)));
1492: info->nz_used = irecv[0];
1493: info->nz_allocated = irecv[1];
1494: info->nz_unneeded = irecv[2];
1495: info->memory = irecv[3];
1496: info->mallocs = irecv[4];
1497: } else SETERRQ(PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Unknown MatInfoType argument %d", (int)flag);
1498: info->fill_ratio_given = 0; /* no parallel LU/ILU/Cholesky */
1499: info->fill_ratio_needed = 0;
1500: info->factor_mallocs = 0;
1501: PetscFunctionReturn(PETSC_SUCCESS);
1502: }
1504: static PetscErrorCode MatSetOption_MPISBAIJ(Mat A, MatOption op, PetscBool flg)
1505: {
1506: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1507: Mat_SeqSBAIJ *aA = (Mat_SeqSBAIJ *)a->A->data;
1509: PetscFunctionBegin;
1510: switch (op) {
1511: case MAT_NEW_NONZERO_LOCATIONS:
1512: case MAT_NEW_NONZERO_ALLOCATION_ERR:
1513: case MAT_UNUSED_NONZERO_LOCATION_ERR:
1514: case MAT_KEEP_NONZERO_PATTERN:
1515: case MAT_SUBMAT_SINGLEIS:
1516: case MAT_NEW_NONZERO_LOCATION_ERR:
1517: MatCheckPreallocated(A, 1);
1518: PetscCall(MatSetOption(a->A, op, flg));
1519: PetscCall(MatSetOption(a->B, op, flg));
1520: break;
1521: case MAT_ROW_ORIENTED:
1522: MatCheckPreallocated(A, 1);
1523: a->roworiented = flg;
1525: PetscCall(MatSetOption(a->A, op, flg));
1526: PetscCall(MatSetOption(a->B, op, flg));
1527: break;
1528: case MAT_FORCE_DIAGONAL_ENTRIES:
1529: case MAT_SORTED_FULL:
1530: PetscCall(PetscInfo(A, "Option %s ignored\n", MatOptions[op]));
1531: break;
1532: case MAT_IGNORE_OFF_PROC_ENTRIES:
1533: a->donotstash = flg;
1534: break;
1535: case MAT_USE_HASH_TABLE:
1536: a->ht_flag = flg;
1537: break;
1538: case MAT_HERMITIAN:
1539: MatCheckPreallocated(A, 1);
1540: PetscCall(MatSetOption(a->A, op, flg));
1541: #if defined(PETSC_USE_COMPLEX)
1542: if (flg) { /* need different mat-vec ops */
1543: A->ops->mult = MatMult_MPISBAIJ_Hermitian;
1544: A->ops->multadd = MatMultAdd_MPISBAIJ_Hermitian;
1545: A->ops->multtranspose = NULL;
1546: A->ops->multtransposeadd = NULL;
1547: A->symmetric = PETSC_BOOL3_FALSE;
1548: }
1549: #endif
1550: break;
1551: case MAT_SPD:
1552: case MAT_SYMMETRIC:
1553: MatCheckPreallocated(A, 1);
1554: PetscCall(MatSetOption(a->A, op, flg));
1555: #if defined(PETSC_USE_COMPLEX)
1556: if (flg) { /* restore to use default mat-vec ops */
1557: A->ops->mult = MatMult_MPISBAIJ;
1558: A->ops->multadd = MatMultAdd_MPISBAIJ;
1559: A->ops->multtranspose = MatMult_MPISBAIJ;
1560: A->ops->multtransposeadd = MatMultAdd_MPISBAIJ;
1561: }
1562: #endif
1563: break;
1564: case MAT_STRUCTURALLY_SYMMETRIC:
1565: MatCheckPreallocated(A, 1);
1566: PetscCall(MatSetOption(a->A, op, flg));
1567: break;
1568: case MAT_SYMMETRY_ETERNAL:
1569: case MAT_STRUCTURAL_SYMMETRY_ETERNAL:
1570: PetscCheck(flg, PETSC_COMM_SELF, PETSC_ERR_SUP, "Matrix must be symmetric");
1571: PetscCall(PetscInfo(A, "Option %s ignored\n", MatOptions[op]));
1572: break;
1573: case MAT_SPD_ETERNAL:
1574: break;
1575: case MAT_IGNORE_LOWER_TRIANGULAR:
1576: aA->ignore_ltriangular = flg;
1577: break;
1578: case MAT_ERROR_LOWER_TRIANGULAR:
1579: aA->ignore_ltriangular = flg;
1580: break;
1581: case MAT_GETROW_UPPERTRIANGULAR:
1582: aA->getrow_utriangular = flg;
1583: break;
1584: default:
1585: SETERRQ(PETSC_COMM_SELF, PETSC_ERR_SUP, "unknown option %d", op);
1586: }
1587: PetscFunctionReturn(PETSC_SUCCESS);
1588: }
1590: static PetscErrorCode MatTranspose_MPISBAIJ(Mat A, MatReuse reuse, Mat *B)
1591: {
1592: PetscFunctionBegin;
1593: if (reuse == MAT_REUSE_MATRIX) PetscCall(MatTransposeCheckNonzeroState_Private(A, *B));
1594: if (reuse == MAT_INITIAL_MATRIX) {
1595: PetscCall(MatDuplicate(A, MAT_COPY_VALUES, B));
1596: } else if (reuse == MAT_REUSE_MATRIX) {
1597: PetscCall(MatCopy(A, *B, SAME_NONZERO_PATTERN));
1598: }
1599: PetscFunctionReturn(PETSC_SUCCESS);
1600: }
1602: static PetscErrorCode MatDiagonalScale_MPISBAIJ(Mat mat, Vec ll, Vec rr)
1603: {
1604: Mat_MPISBAIJ *baij = (Mat_MPISBAIJ *)mat->data;
1605: Mat a = baij->A, b = baij->B;
1606: PetscInt nv, m, n;
1607: PetscBool flg;
1609: PetscFunctionBegin;
1610: if (ll != rr) {
1611: PetscCall(VecEqual(ll, rr, &flg));
1612: PetscCheck(flg, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "For symmetric format, left and right scaling vectors must be same");
1613: }
1614: if (!ll) PetscFunctionReturn(PETSC_SUCCESS);
1616: PetscCall(MatGetLocalSize(mat, &m, &n));
1617: PetscCheck(m == n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "For symmetric format, local size %" PetscInt_FMT " %" PetscInt_FMT " must be same", m, n);
1619: PetscCall(VecGetLocalSize(rr, &nv));
1620: PetscCheck(nv == n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Left and right vector non-conforming local size");
1622: PetscCall(VecScatterBegin(baij->Mvctx, rr, baij->lvec, INSERT_VALUES, SCATTER_FORWARD));
1624: /* left diagonalscale the off-diagonal part */
1625: PetscUseTypeMethod(b, diagonalscale, ll, NULL);
1627: /* scale the diagonal part */
1628: PetscUseTypeMethod(a, diagonalscale, ll, rr);
1630: /* right diagonalscale the off-diagonal part */
1631: PetscCall(VecScatterEnd(baij->Mvctx, rr, baij->lvec, INSERT_VALUES, SCATTER_FORWARD));
1632: PetscUseTypeMethod(b, diagonalscale, NULL, baij->lvec);
1633: PetscFunctionReturn(PETSC_SUCCESS);
1634: }
1636: static PetscErrorCode MatSetUnfactored_MPISBAIJ(Mat A)
1637: {
1638: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1640: PetscFunctionBegin;
1641: PetscCall(MatSetUnfactored(a->A));
1642: PetscFunctionReturn(PETSC_SUCCESS);
1643: }
1645: static PetscErrorCode MatDuplicate_MPISBAIJ(Mat, MatDuplicateOption, Mat *);
1647: static PetscErrorCode MatEqual_MPISBAIJ(Mat A, Mat B, PetscBool *flag)
1648: {
1649: Mat_MPISBAIJ *matB = (Mat_MPISBAIJ *)B->data, *matA = (Mat_MPISBAIJ *)A->data;
1650: Mat a, b, c, d;
1651: PetscBool flg;
1653: PetscFunctionBegin;
1654: a = matA->A;
1655: b = matA->B;
1656: c = matB->A;
1657: d = matB->B;
1659: PetscCall(MatEqual(a, c, &flg));
1660: if (flg) PetscCall(MatEqual(b, d, &flg));
1661: PetscCallMPI(MPIU_Allreduce(&flg, flag, 1, MPIU_BOOL, MPI_LAND, PetscObjectComm((PetscObject)A)));
1662: PetscFunctionReturn(PETSC_SUCCESS);
1663: }
1665: static PetscErrorCode MatCopy_MPISBAIJ(Mat A, Mat B, MatStructure str)
1666: {
1667: PetscBool isbaij;
1669: PetscFunctionBegin;
1670: PetscCall(PetscObjectTypeCompareAny((PetscObject)B, &isbaij, MATSEQSBAIJ, MATMPISBAIJ, ""));
1671: PetscCheck(isbaij, PetscObjectComm((PetscObject)B), PETSC_ERR_SUP, "Not for matrix type %s", ((PetscObject)B)->type_name);
1672: /* If the two matrices don't have the same copy implementation, they aren't compatible for fast copy. */
1673: if ((str != SAME_NONZERO_PATTERN) || (A->ops->copy != B->ops->copy)) {
1674: PetscCall(MatGetRowUpperTriangular(A));
1675: PetscCall(MatCopy_Basic(A, B, str));
1676: PetscCall(MatRestoreRowUpperTriangular(A));
1677: } else {
1678: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1679: Mat_MPISBAIJ *b = (Mat_MPISBAIJ *)B->data;
1681: PetscCall(MatCopy(a->A, b->A, str));
1682: PetscCall(MatCopy(a->B, b->B, str));
1683: }
1684: PetscCall(PetscObjectStateIncrease((PetscObject)B));
1685: PetscFunctionReturn(PETSC_SUCCESS);
1686: }
1688: static PetscErrorCode MatAXPY_MPISBAIJ(Mat Y, PetscScalar a, Mat X, MatStructure str)
1689: {
1690: Mat_MPISBAIJ *xx = (Mat_MPISBAIJ *)X->data, *yy = (Mat_MPISBAIJ *)Y->data;
1691: PetscBLASInt bnz, one = 1;
1692: Mat_SeqSBAIJ *xa, *ya;
1693: Mat_SeqBAIJ *xb, *yb;
1695: PetscFunctionBegin;
1696: if (str == SAME_NONZERO_PATTERN) {
1697: PetscScalar alpha = a;
1698: xa = (Mat_SeqSBAIJ *)xx->A->data;
1699: ya = (Mat_SeqSBAIJ *)yy->A->data;
1700: PetscCall(PetscBLASIntCast(xa->nz, &bnz));
1701: PetscCallBLAS("BLASaxpy", BLASaxpy_(&bnz, &alpha, xa->a, &one, ya->a, &one));
1702: xb = (Mat_SeqBAIJ *)xx->B->data;
1703: yb = (Mat_SeqBAIJ *)yy->B->data;
1704: PetscCall(PetscBLASIntCast(xb->nz, &bnz));
1705: PetscCallBLAS("BLASaxpy", BLASaxpy_(&bnz, &alpha, xb->a, &one, yb->a, &one));
1706: PetscCall(PetscObjectStateIncrease((PetscObject)Y));
1707: } else if (str == SUBSET_NONZERO_PATTERN) { /* nonzeros of X is a subset of Y's */
1708: PetscCall(MatSetOption(X, MAT_GETROW_UPPERTRIANGULAR, PETSC_TRUE));
1709: PetscCall(MatAXPY_Basic(Y, a, X, str));
1710: PetscCall(MatSetOption(X, MAT_GETROW_UPPERTRIANGULAR, PETSC_FALSE));
1711: } else {
1712: Mat B;
1713: PetscInt *nnz_d, *nnz_o, bs = Y->rmap->bs;
1714: PetscCheck(bs == X->rmap->bs, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "Matrices must have same block size");
1715: PetscCall(MatGetRowUpperTriangular(X));
1716: PetscCall(MatGetRowUpperTriangular(Y));
1717: PetscCall(PetscMalloc1(yy->A->rmap->N, &nnz_d));
1718: PetscCall(PetscMalloc1(yy->B->rmap->N, &nnz_o));
1719: PetscCall(MatCreate(PetscObjectComm((PetscObject)Y), &B));
1720: PetscCall(PetscObjectSetName((PetscObject)B, ((PetscObject)Y)->name));
1721: PetscCall(MatSetSizes(B, Y->rmap->n, Y->cmap->n, Y->rmap->N, Y->cmap->N));
1722: PetscCall(MatSetBlockSizesFromMats(B, Y, Y));
1723: PetscCall(MatSetType(B, MATMPISBAIJ));
1724: PetscCall(MatAXPYGetPreallocation_SeqSBAIJ(yy->A, xx->A, nnz_d));
1725: PetscCall(MatAXPYGetPreallocation_MPIBAIJ(yy->B, yy->garray, xx->B, xx->garray, nnz_o));
1726: PetscCall(MatMPISBAIJSetPreallocation(B, bs, 0, nnz_d, 0, nnz_o));
1727: PetscCall(MatAXPY_BasicWithPreallocation(B, Y, a, X, str));
1728: PetscCall(MatHeaderMerge(Y, &B));
1729: PetscCall(PetscFree(nnz_d));
1730: PetscCall(PetscFree(nnz_o));
1731: PetscCall(MatRestoreRowUpperTriangular(X));
1732: PetscCall(MatRestoreRowUpperTriangular(Y));
1733: }
1734: PetscFunctionReturn(PETSC_SUCCESS);
1735: }
1737: static PetscErrorCode MatCreateSubMatrices_MPISBAIJ(Mat A, PetscInt n, const IS irow[], const IS icol[], MatReuse scall, Mat *B[])
1738: {
1739: PetscInt i;
1740: PetscBool flg;
1742: PetscFunctionBegin;
1743: PetscCall(MatCreateSubMatrices_MPIBAIJ(A, n, irow, icol, scall, B)); /* B[] are sbaij matrices */
1744: for (i = 0; i < n; i++) {
1745: PetscCall(ISEqual(irow[i], icol[i], &flg));
1746: if (!flg) PetscCall(MatSeqSBAIJZeroOps_Private(*B[i]));
1747: }
1748: PetscFunctionReturn(PETSC_SUCCESS);
1749: }
1751: static PetscErrorCode MatShift_MPISBAIJ(Mat Y, PetscScalar a)
1752: {
1753: Mat_MPISBAIJ *maij = (Mat_MPISBAIJ *)Y->data;
1754: Mat_SeqSBAIJ *aij = (Mat_SeqSBAIJ *)maij->A->data;
1756: PetscFunctionBegin;
1757: if (!Y->preallocated) {
1758: PetscCall(MatMPISBAIJSetPreallocation(Y, Y->rmap->bs, 1, NULL, 0, NULL));
1759: } else if (!aij->nz) {
1760: PetscInt nonew = aij->nonew;
1761: PetscCall(MatSeqSBAIJSetPreallocation(maij->A, Y->rmap->bs, 1, NULL));
1762: aij->nonew = nonew;
1763: }
1764: PetscCall(MatShift_Basic(Y, a));
1765: PetscFunctionReturn(PETSC_SUCCESS);
1766: }
1768: static PetscErrorCode MatMissingDiagonal_MPISBAIJ(Mat A, PetscBool *missing, PetscInt *d)
1769: {
1770: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1772: PetscFunctionBegin;
1773: PetscCheck(A->rmap->n == A->cmap->n, PETSC_COMM_SELF, PETSC_ERR_SUP, "Only works for square matrices");
1774: PetscCall(MatMissingDiagonal(a->A, missing, d));
1775: if (d) {
1776: PetscInt rstart;
1777: PetscCall(MatGetOwnershipRange(A, &rstart, NULL));
1778: *d += rstart / A->rmap->bs;
1779: }
1780: PetscFunctionReturn(PETSC_SUCCESS);
1781: }
1783: static PetscErrorCode MatGetDiagonalBlock_MPISBAIJ(Mat A, Mat *a)
1784: {
1785: PetscFunctionBegin;
1786: *a = ((Mat_MPISBAIJ *)A->data)->A;
1787: PetscFunctionReturn(PETSC_SUCCESS);
1788: }
1790: static PetscErrorCode MatEliminateZeros_MPISBAIJ(Mat A, PetscBool keep)
1791: {
1792: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
1794: PetscFunctionBegin;
1795: PetscCall(MatEliminateZeros_SeqSBAIJ(a->A, keep)); // possibly keep zero diagonal coefficients
1796: PetscCall(MatEliminateZeros_SeqBAIJ(a->B, PETSC_FALSE)); // never keep zero diagonal coefficients
1797: PetscFunctionReturn(PETSC_SUCCESS);
1798: }
1800: static PetscErrorCode MatLoad_MPISBAIJ(Mat, PetscViewer);
1801: static PetscErrorCode MatGetRowMaxAbs_MPISBAIJ(Mat, Vec, PetscInt[]);
1802: static PetscErrorCode MatSOR_MPISBAIJ(Mat, Vec, PetscReal, MatSORType, PetscReal, PetscInt, PetscInt, Vec);
1804: static struct _MatOps MatOps_Values = {MatSetValues_MPISBAIJ,
1805: MatGetRow_MPISBAIJ,
1806: MatRestoreRow_MPISBAIJ,
1807: MatMult_MPISBAIJ,
1808: /* 4*/ MatMultAdd_MPISBAIJ,
1809: MatMult_MPISBAIJ, /* transpose versions are same as non-transpose */
1810: MatMultAdd_MPISBAIJ,
1811: NULL,
1812: NULL,
1813: NULL,
1814: /* 10*/ NULL,
1815: NULL,
1816: NULL,
1817: MatSOR_MPISBAIJ,
1818: MatTranspose_MPISBAIJ,
1819: /* 15*/ MatGetInfo_MPISBAIJ,
1820: MatEqual_MPISBAIJ,
1821: MatGetDiagonal_MPISBAIJ,
1822: MatDiagonalScale_MPISBAIJ,
1823: MatNorm_MPISBAIJ,
1824: /* 20*/ MatAssemblyBegin_MPISBAIJ,
1825: MatAssemblyEnd_MPISBAIJ,
1826: MatSetOption_MPISBAIJ,
1827: MatZeroEntries_MPISBAIJ,
1828: /* 24*/ NULL,
1829: NULL,
1830: NULL,
1831: NULL,
1832: NULL,
1833: /* 29*/ MatSetUp_MPI_Hash,
1834: NULL,
1835: NULL,
1836: MatGetDiagonalBlock_MPISBAIJ,
1837: NULL,
1838: /* 34*/ MatDuplicate_MPISBAIJ,
1839: NULL,
1840: NULL,
1841: NULL,
1842: NULL,
1843: /* 39*/ MatAXPY_MPISBAIJ,
1844: MatCreateSubMatrices_MPISBAIJ,
1845: MatIncreaseOverlap_MPISBAIJ,
1846: MatGetValues_MPISBAIJ,
1847: MatCopy_MPISBAIJ,
1848: /* 44*/ NULL,
1849: MatScale_MPISBAIJ,
1850: MatShift_MPISBAIJ,
1851: NULL,
1852: NULL,
1853: /* 49*/ NULL,
1854: NULL,
1855: NULL,
1856: NULL,
1857: NULL,
1858: /* 54*/ NULL,
1859: NULL,
1860: MatSetUnfactored_MPISBAIJ,
1861: NULL,
1862: MatSetValuesBlocked_MPISBAIJ,
1863: /* 59*/ MatCreateSubMatrix_MPISBAIJ,
1864: NULL,
1865: NULL,
1866: NULL,
1867: NULL,
1868: /* 64*/ NULL,
1869: NULL,
1870: NULL,
1871: NULL,
1872: NULL,
1873: /* 69*/ MatGetRowMaxAbs_MPISBAIJ,
1874: NULL,
1875: MatConvert_MPISBAIJ_Basic,
1876: NULL,
1877: NULL,
1878: /* 74*/ NULL,
1879: NULL,
1880: NULL,
1881: NULL,
1882: NULL,
1883: /* 79*/ NULL,
1884: NULL,
1885: NULL,
1886: NULL,
1887: MatLoad_MPISBAIJ,
1888: /* 84*/ NULL,
1889: NULL,
1890: NULL,
1891: NULL,
1892: NULL,
1893: /* 89*/ NULL,
1894: NULL,
1895: NULL,
1896: NULL,
1897: NULL,
1898: /* 94*/ NULL,
1899: NULL,
1900: NULL,
1901: NULL,
1902: NULL,
1903: /* 99*/ NULL,
1904: NULL,
1905: NULL,
1906: MatConjugate_MPISBAIJ,
1907: NULL,
1908: /*104*/ NULL,
1909: MatRealPart_MPISBAIJ,
1910: MatImaginaryPart_MPISBAIJ,
1911: MatGetRowUpperTriangular_MPISBAIJ,
1912: MatRestoreRowUpperTriangular_MPISBAIJ,
1913: /*109*/ NULL,
1914: NULL,
1915: NULL,
1916: NULL,
1917: MatMissingDiagonal_MPISBAIJ,
1918: /*114*/ NULL,
1919: NULL,
1920: NULL,
1921: NULL,
1922: NULL,
1923: /*119*/ NULL,
1924: NULL,
1925: NULL,
1926: NULL,
1927: NULL,
1928: /*124*/ NULL,
1929: NULL,
1930: NULL,
1931: NULL,
1932: NULL,
1933: /*129*/ NULL,
1934: NULL,
1935: NULL,
1936: NULL,
1937: NULL,
1938: /*134*/ NULL,
1939: NULL,
1940: NULL,
1941: NULL,
1942: NULL,
1943: /*139*/ MatSetBlockSizes_Default,
1944: NULL,
1945: NULL,
1946: NULL,
1947: NULL,
1948: /*144*/ MatCreateMPIMatConcatenateSeqMat_MPISBAIJ,
1949: NULL,
1950: NULL,
1951: NULL,
1952: NULL,
1953: NULL,
1954: /*150*/ NULL,
1955: MatEliminateZeros_MPISBAIJ,
1956: NULL,
1957: NULL,
1958: NULL,
1959: /*155*/ NULL,
1960: MatCopyHashToXAIJ_MPI_Hash};
1962: static PetscErrorCode MatMPISBAIJSetPreallocation_MPISBAIJ(Mat B, PetscInt bs, PetscInt d_nz, const PetscInt *d_nnz, PetscInt o_nz, const PetscInt *o_nnz)
1963: {
1964: Mat_MPISBAIJ *b = (Mat_MPISBAIJ *)B->data;
1965: PetscInt i, mbs, Mbs;
1966: PetscMPIInt size;
1968: PetscFunctionBegin;
1969: if (B->hash_active) {
1970: B->ops[0] = b->cops;
1971: B->hash_active = PETSC_FALSE;
1972: }
1973: if (!B->preallocated) PetscCall(MatStashCreate_Private(PetscObjectComm((PetscObject)B), bs, &B->bstash));
1974: PetscCall(MatSetBlockSize(B, PetscAbs(bs)));
1975: PetscCall(PetscLayoutSetUp(B->rmap));
1976: PetscCall(PetscLayoutSetUp(B->cmap));
1977: PetscCall(PetscLayoutGetBlockSize(B->rmap, &bs));
1978: PetscCheck(B->rmap->N <= B->cmap->N, PetscObjectComm((PetscObject)B), PETSC_ERR_SUP, "MPISBAIJ matrix cannot have more rows %" PetscInt_FMT " than columns %" PetscInt_FMT, B->rmap->N, B->cmap->N);
1979: PetscCheck(B->rmap->n <= B->cmap->n, PETSC_COMM_SELF, PETSC_ERR_SUP, "MPISBAIJ matrix cannot have more local rows %" PetscInt_FMT " than columns %" PetscInt_FMT, B->rmap->n, B->cmap->n);
1981: mbs = B->rmap->n / bs;
1982: Mbs = B->rmap->N / bs;
1983: PetscCheck(mbs * bs == B->rmap->n, PETSC_COMM_SELF, PETSC_ERR_ARG_SIZ, "No of local rows %" PetscInt_FMT " must be divisible by blocksize %" PetscInt_FMT, B->rmap->N, bs);
1985: B->rmap->bs = bs;
1986: b->bs2 = bs * bs;
1987: b->mbs = mbs;
1988: b->Mbs = Mbs;
1989: b->nbs = B->cmap->n / bs;
1990: b->Nbs = B->cmap->N / bs;
1992: for (i = 0; i <= b->size; i++) b->rangebs[i] = B->rmap->range[i] / bs;
1993: b->rstartbs = B->rmap->rstart / bs;
1994: b->rendbs = B->rmap->rend / bs;
1996: b->cstartbs = B->cmap->rstart / bs;
1997: b->cendbs = B->cmap->rend / bs;
1999: #if defined(PETSC_USE_CTABLE)
2000: PetscCall(PetscHMapIDestroy(&b->colmap));
2001: #else
2002: PetscCall(PetscFree(b->colmap));
2003: #endif
2004: PetscCall(PetscFree(b->garray));
2005: PetscCall(VecDestroy(&b->lvec));
2006: PetscCall(VecScatterDestroy(&b->Mvctx));
2007: PetscCall(VecDestroy(&b->slvec0));
2008: PetscCall(VecDestroy(&b->slvec0b));
2009: PetscCall(VecDestroy(&b->slvec1));
2010: PetscCall(VecDestroy(&b->slvec1a));
2011: PetscCall(VecDestroy(&b->slvec1b));
2012: PetscCall(VecScatterDestroy(&b->sMvctx));
2014: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)B), &size));
2016: MatSeqXAIJGetOptions_Private(b->B);
2017: PetscCall(MatDestroy(&b->B));
2018: PetscCall(MatCreate(PETSC_COMM_SELF, &b->B));
2019: PetscCall(MatSetSizes(b->B, B->rmap->n, size > 1 ? B->cmap->N : 0, B->rmap->n, size > 1 ? B->cmap->N : 0));
2020: PetscCall(MatSetType(b->B, MATSEQBAIJ));
2021: MatSeqXAIJRestoreOptions_Private(b->B);
2023: MatSeqXAIJGetOptions_Private(b->A);
2024: PetscCall(MatDestroy(&b->A));
2025: PetscCall(MatCreate(PETSC_COMM_SELF, &b->A));
2026: PetscCall(MatSetSizes(b->A, B->rmap->n, B->cmap->n, B->rmap->n, B->cmap->n));
2027: PetscCall(MatSetType(b->A, MATSEQSBAIJ));
2028: MatSeqXAIJRestoreOptions_Private(b->A);
2030: PetscCall(MatSeqSBAIJSetPreallocation(b->A, bs, d_nz, d_nnz));
2031: PetscCall(MatSeqBAIJSetPreallocation(b->B, bs, o_nz, o_nnz));
2033: B->preallocated = PETSC_TRUE;
2034: B->was_assembled = PETSC_FALSE;
2035: B->assembled = PETSC_FALSE;
2036: PetscFunctionReturn(PETSC_SUCCESS);
2037: }
2039: static PetscErrorCode MatMPISBAIJSetPreallocationCSR_MPISBAIJ(Mat B, PetscInt bs, const PetscInt ii[], const PetscInt jj[], const PetscScalar V[])
2040: {
2041: PetscInt m, rstart, cend;
2042: PetscInt i, j, d, nz, bd, nz_max = 0, *d_nnz = NULL, *o_nnz = NULL;
2043: const PetscInt *JJ = NULL;
2044: PetscScalar *values = NULL;
2045: PetscBool roworiented = ((Mat_MPISBAIJ *)B->data)->roworiented;
2046: PetscBool nooffprocentries;
2048: PetscFunctionBegin;
2049: PetscCheck(bs >= 1, PetscObjectComm((PetscObject)B), PETSC_ERR_ARG_OUTOFRANGE, "Invalid block size specified, must be positive but it is %" PetscInt_FMT, bs);
2050: PetscCall(PetscLayoutSetBlockSize(B->rmap, bs));
2051: PetscCall(PetscLayoutSetBlockSize(B->cmap, bs));
2052: PetscCall(PetscLayoutSetUp(B->rmap));
2053: PetscCall(PetscLayoutSetUp(B->cmap));
2054: PetscCall(PetscLayoutGetBlockSize(B->rmap, &bs));
2055: m = B->rmap->n / bs;
2056: rstart = B->rmap->rstart / bs;
2057: cend = B->cmap->rend / bs;
2059: PetscCheck(!ii[0], PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "ii[0] must be 0 but it is %" PetscInt_FMT, ii[0]);
2060: PetscCall(PetscMalloc2(m, &d_nnz, m, &o_nnz));
2061: for (i = 0; i < m; i++) {
2062: nz = ii[i + 1] - ii[i];
2063: PetscCheck(nz >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "Local row %" PetscInt_FMT " has a negative number of columns %" PetscInt_FMT, i, nz);
2064: /* count the ones on the diagonal and above, split into diagonal and off-diagonal portions. */
2065: JJ = jj + ii[i];
2066: bd = 0;
2067: for (j = 0; j < nz; j++) {
2068: if (*JJ >= i + rstart) break;
2069: JJ++;
2070: bd++;
2071: }
2072: d = 0;
2073: for (; j < nz; j++) {
2074: if (*JJ++ >= cend) break;
2075: d++;
2076: }
2077: d_nnz[i] = d;
2078: o_nnz[i] = nz - d - bd;
2079: nz = nz - bd;
2080: nz_max = PetscMax(nz_max, nz);
2081: }
2082: PetscCall(MatMPISBAIJSetPreallocation(B, bs, 0, d_nnz, 0, o_nnz));
2083: PetscCall(MatSetOption(B, MAT_IGNORE_LOWER_TRIANGULAR, PETSC_TRUE));
2084: PetscCall(PetscFree2(d_nnz, o_nnz));
2086: values = (PetscScalar *)V;
2087: if (!values) PetscCall(PetscCalloc1(bs * bs * nz_max, &values));
2088: for (i = 0; i < m; i++) {
2089: PetscInt row = i + rstart;
2090: PetscInt ncols = ii[i + 1] - ii[i];
2091: const PetscInt *icols = jj + ii[i];
2092: if (bs == 1 || !roworiented) { /* block ordering matches the non-nested layout of MatSetValues so we can insert entire rows */
2093: const PetscScalar *svals = values + (V ? (bs * bs * ii[i]) : 0);
2094: PetscCall(MatSetValuesBlocked_MPISBAIJ(B, 1, &row, ncols, icols, svals, INSERT_VALUES));
2095: } else { /* block ordering does not match so we can only insert one block at a time. */
2096: PetscInt j;
2097: for (j = 0; j < ncols; j++) {
2098: const PetscScalar *svals = values + (V ? (bs * bs * (ii[i] + j)) : 0);
2099: PetscCall(MatSetValuesBlocked_MPISBAIJ(B, 1, &row, 1, &icols[j], svals, INSERT_VALUES));
2100: }
2101: }
2102: }
2104: if (!V) PetscCall(PetscFree(values));
2105: nooffprocentries = B->nooffprocentries;
2106: B->nooffprocentries = PETSC_TRUE;
2107: PetscCall(MatAssemblyBegin(B, MAT_FINAL_ASSEMBLY));
2108: PetscCall(MatAssemblyEnd(B, MAT_FINAL_ASSEMBLY));
2109: B->nooffprocentries = nooffprocentries;
2111: PetscCall(MatSetOption(B, MAT_NEW_NONZERO_LOCATION_ERR, PETSC_TRUE));
2112: PetscFunctionReturn(PETSC_SUCCESS);
2113: }
2115: /*MC
2116: MATMPISBAIJ - MATMPISBAIJ = "mpisbaij" - A matrix type to be used for distributed symmetric sparse block matrices,
2117: based on block compressed sparse row format. Only the upper triangular portion of the "diagonal" portion of
2118: the matrix is stored.
2120: For complex numbers by default this matrix is symmetric, NOT Hermitian symmetric. To make it Hermitian symmetric you
2121: can call `MatSetOption`(`Mat`, `MAT_HERMITIAN`);
2123: Options Database Key:
2124: . -mat_type mpisbaij - sets the matrix type to "mpisbaij" during a call to `MatSetFromOptions()`
2126: Level: beginner
2128: Note:
2129: The number of rows in the matrix must be less than or equal to the number of columns. Similarly the number of rows in the
2130: diagonal portion of the matrix of each process has to less than or equal the number of columns.
2132: .seealso: [](ch_matrices), `Mat`, `MATSBAIJ`, `MATBAIJ`, `MatCreateBAIJ()`, `MATSEQSBAIJ`, `MatType`
2133: M*/
2135: PETSC_EXTERN PetscErrorCode MatCreate_MPISBAIJ(Mat B)
2136: {
2137: Mat_MPISBAIJ *b;
2138: PetscBool flg = PETSC_FALSE;
2140: PetscFunctionBegin;
2141: PetscCall(PetscNew(&b));
2142: B->data = (void *)b;
2143: B->ops[0] = MatOps_Values;
2145: B->ops->destroy = MatDestroy_MPISBAIJ;
2146: B->ops->view = MatView_MPISBAIJ;
2147: B->assembled = PETSC_FALSE;
2148: B->insertmode = NOT_SET_VALUES;
2150: PetscCallMPI(MPI_Comm_rank(PetscObjectComm((PetscObject)B), &b->rank));
2151: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)B), &b->size));
2153: /* build local table of row and column ownerships */
2154: PetscCall(PetscMalloc1(b->size + 2, &b->rangebs));
2156: /* build cache for off array entries formed */
2157: PetscCall(MatStashCreate_Private(PetscObjectComm((PetscObject)B), 1, &B->stash));
2159: b->donotstash = PETSC_FALSE;
2160: b->colmap = NULL;
2161: b->garray = NULL;
2162: b->roworiented = PETSC_TRUE;
2164: /* stuff used in block assembly */
2165: b->barray = NULL;
2167: /* stuff used for matrix vector multiply */
2168: b->lvec = NULL;
2169: b->Mvctx = NULL;
2170: b->slvec0 = NULL;
2171: b->slvec0b = NULL;
2172: b->slvec1 = NULL;
2173: b->slvec1a = NULL;
2174: b->slvec1b = NULL;
2175: b->sMvctx = NULL;
2177: /* stuff for MatGetRow() */
2178: b->rowindices = NULL;
2179: b->rowvalues = NULL;
2180: b->getrowactive = PETSC_FALSE;
2182: /* hash table stuff */
2183: b->ht = NULL;
2184: b->hd = NULL;
2185: b->ht_size = 0;
2186: b->ht_flag = PETSC_FALSE;
2187: b->ht_fact = 0;
2188: b->ht_total_ct = 0;
2189: b->ht_insert_ct = 0;
2191: /* stuff for MatCreateSubMatrices_MPIBAIJ_local() */
2192: b->ijonly = PETSC_FALSE;
2194: b->in_loc = NULL;
2195: b->v_loc = NULL;
2196: b->n_loc = 0;
2198: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatStoreValues_C", MatStoreValues_MPISBAIJ));
2199: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatRetrieveValues_C", MatRetrieveValues_MPISBAIJ));
2200: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatMPISBAIJSetPreallocation_C", MatMPISBAIJSetPreallocation_MPISBAIJ));
2201: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatMPISBAIJSetPreallocationCSR_C", MatMPISBAIJSetPreallocationCSR_MPISBAIJ));
2202: #if defined(PETSC_HAVE_ELEMENTAL)
2203: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_mpisbaij_elemental_C", MatConvert_MPISBAIJ_Elemental));
2204: #endif
2205: #if defined(PETSC_HAVE_SCALAPACK)
2206: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_mpisbaij_scalapack_C", MatConvert_SBAIJ_ScaLAPACK));
2207: #endif
2208: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_mpisbaij_mpiaij_C", MatConvert_MPISBAIJ_Basic));
2209: PetscCall(PetscObjectComposeFunction((PetscObject)B, "MatConvert_mpisbaij_mpibaij_C", MatConvert_MPISBAIJ_Basic));
2211: B->symmetric = PETSC_BOOL3_TRUE;
2212: B->structurally_symmetric = PETSC_BOOL3_TRUE;
2213: B->symmetry_eternal = PETSC_TRUE;
2214: B->structural_symmetry_eternal = PETSC_TRUE;
2215: #if defined(PETSC_USE_COMPLEX)
2216: B->hermitian = PETSC_BOOL3_FALSE;
2217: #else
2218: B->hermitian = PETSC_BOOL3_TRUE;
2219: #endif
2221: PetscCall(PetscObjectChangeTypeName((PetscObject)B, MATMPISBAIJ));
2222: PetscOptionsBegin(PetscObjectComm((PetscObject)B), NULL, "Options for loading MPISBAIJ matrix 1", "Mat");
2223: PetscCall(PetscOptionsBool("-mat_use_hash_table", "Use hash table to save memory in constructing matrix", "MatSetOption", flg, &flg, NULL));
2224: if (flg) {
2225: PetscReal fact = 1.39;
2226: PetscCall(MatSetOption(B, MAT_USE_HASH_TABLE, PETSC_TRUE));
2227: PetscCall(PetscOptionsReal("-mat_use_hash_table", "Use hash table factor", "MatMPIBAIJSetHashTableFactor", fact, &fact, NULL));
2228: if (fact <= 1.0) fact = 1.39;
2229: PetscCall(MatMPIBAIJSetHashTableFactor(B, fact));
2230: PetscCall(PetscInfo(B, "Hash table Factor used %5.2g\n", (double)fact));
2231: }
2232: PetscOptionsEnd();
2233: PetscFunctionReturn(PETSC_SUCCESS);
2234: }
2236: // PetscClangLinter pragma disable: -fdoc-section-header-unknown
2237: /*MC
2238: MATSBAIJ - MATSBAIJ = "sbaij" - A matrix type to be used for symmetric block sparse matrices.
2240: This matrix type is identical to `MATSEQSBAIJ` when constructed with a single process communicator,
2241: and `MATMPISBAIJ` otherwise.
2243: Options Database Key:
2244: . -mat_type sbaij - sets the matrix type to `MATSBAIJ` during a call to `MatSetFromOptions()`
2246: Level: beginner
2248: .seealso: [](ch_matrices), `Mat`, `MATSEQSBAIJ`, `MATMPISBAIJ`, `MatCreateSBAIJ()`, `MATSEQSBAIJ`, `MATMPISBAIJ`
2249: M*/
2251: /*@
2252: MatMPISBAIJSetPreallocation - For good matrix assembly performance
2253: the user should preallocate the matrix storage by setting the parameters
2254: d_nz (or d_nnz) and o_nz (or o_nnz). By setting these parameters accurately,
2255: performance can be increased by more than a factor of 50.
2257: Collective
2259: Input Parameters:
2260: + B - the matrix
2261: . bs - size of block, the blocks are ALWAYS square. One can use MatSetBlockSizes() to set a different row and column blocksize but the row
2262: blocksize always defines the size of the blocks. The column blocksize sets the blocksize of the vectors obtained with MatCreateVecs()
2263: . d_nz - number of block nonzeros per block row in diagonal portion of local
2264: submatrix (same for all local rows)
2265: . d_nnz - array containing the number of block nonzeros in the various block rows
2266: in the upper triangular and diagonal part of the in diagonal portion of the local
2267: (possibly different for each block row) or `NULL`. If you plan to factor the matrix you must leave room
2268: for the diagonal entry and set a value even if it is zero.
2269: . o_nz - number of block nonzeros per block row in the off-diagonal portion of local
2270: submatrix (same for all local rows).
2271: - o_nnz - array containing the number of nonzeros in the various block rows of the
2272: off-diagonal portion of the local submatrix that is right of the diagonal
2273: (possibly different for each block row) or `NULL`.
2275: Options Database Keys:
2276: + -mat_no_unroll - uses code that does not unroll the loops in the
2277: block calculations (much slower)
2278: - -mat_block_size - size of the blocks to use
2280: Level: intermediate
2282: Notes:
2284: If `PETSC_DECIDE` or `PETSC_DETERMINE` is used for a particular argument on one processor
2285: than it must be used on all processors that share the object for that argument.
2287: If the *_nnz parameter is given then the *_nz parameter is ignored
2289: Storage Information:
2290: For a square global matrix we define each processor's diagonal portion
2291: to be its local rows and the corresponding columns (a square submatrix);
2292: each processor's off-diagonal portion encompasses the remainder of the
2293: local matrix (a rectangular submatrix).
2295: The user can specify preallocated storage for the diagonal part of
2296: the local submatrix with either `d_nz` or `d_nnz` (not both). Set
2297: `d_nz` = `PETSC_DEFAULT` and `d_nnz` = `NULL` for PETSc to control dynamic
2298: memory allocation. Likewise, specify preallocated storage for the
2299: off-diagonal part of the local submatrix with `o_nz` or `o_nnz` (not both).
2301: You can call `MatGetInfo()` to get information on how effective the preallocation was;
2302: for example the fields mallocs,nz_allocated,nz_used,nz_unneeded;
2303: You can also run with the option `-info` and look for messages with the string
2304: malloc in them to see if additional memory allocation was needed.
2306: Consider a processor that owns rows 3, 4 and 5 of a parallel matrix. In
2307: the figure below we depict these three local rows and all columns (0-11).
2309: .vb
2310: 0 1 2 3 4 5 6 7 8 9 10 11
2311: --------------------------
2312: row 3 |. . . d d d o o o o o o
2313: row 4 |. . . d d d o o o o o o
2314: row 5 |. . . d d d o o o o o o
2315: --------------------------
2316: .ve
2318: Thus, any entries in the d locations are stored in the d (diagonal)
2319: submatrix, and any entries in the o locations are stored in the
2320: o (off-diagonal) submatrix. Note that the d matrix is stored in
2321: `MATSEQSBAIJ` format and the o submatrix in `MATSEQBAIJ` format.
2323: Now `d_nz` should indicate the number of block nonzeros per row in the upper triangular
2324: plus the diagonal part of the d matrix,
2325: and `o_nz` should indicate the number of block nonzeros per row in the o matrix
2327: In general, for PDE problems in which most nonzeros are near the diagonal,
2328: one expects `d_nz` >> `o_nz`.
2330: .seealso: [](ch_matrices), `Mat`, `MATMPISBAIJ`, `MATSBAIJ`, `MatCreate()`, `MatCreateSeqSBAIJ()`, `MatSetValues()`, `MatCreateBAIJ()`, `PetscSplitOwnership()`
2331: @*/
2332: PetscErrorCode MatMPISBAIJSetPreallocation(Mat B, PetscInt bs, PetscInt d_nz, const PetscInt d_nnz[], PetscInt o_nz, const PetscInt o_nnz[])
2333: {
2334: PetscFunctionBegin;
2338: PetscTryMethod(B, "MatMPISBAIJSetPreallocation_C", (Mat, PetscInt, PetscInt, const PetscInt[], PetscInt, const PetscInt[]), (B, bs, d_nz, d_nnz, o_nz, o_nnz));
2339: PetscFunctionReturn(PETSC_SUCCESS);
2340: }
2342: // PetscClangLinter pragma disable: -fdoc-section-header-unknown
2343: /*@
2344: MatCreateSBAIJ - Creates a sparse parallel matrix in symmetric block AIJ format, `MATSBAIJ`,
2345: (block compressed row). For good matrix assembly performance
2346: the user should preallocate the matrix storage by setting the parameters
2347: `d_nz` (or `d_nnz`) and `o_nz` (or `o_nnz`).
2349: Collective
2351: Input Parameters:
2352: + comm - MPI communicator
2353: . bs - size of block, the blocks are ALWAYS square. One can use `MatSetBlockSizes()` to set a different row and column blocksize but the row
2354: blocksize always defines the size of the blocks. The column blocksize sets the blocksize of the vectors obtained with `MatCreateVecs()`
2355: . m - number of local rows (or `PETSC_DECIDE` to have calculated if `M` is given)
2356: This value should be the same as the local size used in creating the
2357: y vector for the matrix-vector product y = Ax.
2358: . n - number of local columns (or `PETSC_DECIDE` to have calculated if `N` is given)
2359: This value should be the same as the local size used in creating the
2360: x vector for the matrix-vector product y = Ax.
2361: . M - number of global rows (or `PETSC_DETERMINE` to have calculated if `m` is given)
2362: . N - number of global columns (or `PETSC_DETERMINE` to have calculated if `n` is given)
2363: . d_nz - number of block nonzeros per block row in diagonal portion of local
2364: submatrix (same for all local rows)
2365: . d_nnz - array containing the number of block nonzeros in the various block rows
2366: in the upper triangular portion of the in diagonal portion of the local
2367: (possibly different for each block block row) or `NULL`.
2368: If you plan to factor the matrix you must leave room for the diagonal entry and
2369: set its value even if it is zero.
2370: . o_nz - number of block nonzeros per block row in the off-diagonal portion of local
2371: submatrix (same for all local rows).
2372: - o_nnz - array containing the number of nonzeros in the various block rows of the
2373: off-diagonal portion of the local submatrix (possibly different for
2374: each block row) or `NULL`.
2376: Output Parameter:
2377: . A - the matrix
2379: Options Database Keys:
2380: + -mat_no_unroll - uses code that does not unroll the loops in the
2381: block calculations (much slower)
2382: . -mat_block_size - size of the blocks to use
2383: - -mat_mpi - use the parallel matrix data structures even on one processor
2384: (defaults to using SeqBAIJ format on one processor)
2386: Level: intermediate
2388: Notes:
2389: It is recommended that one use `MatCreateFromOptions()` or the `MatCreate()`, `MatSetType()` and/or `MatSetFromOptions()`,
2390: MatXXXXSetPreallocation() paradigm instead of this routine directly.
2391: [MatXXXXSetPreallocation() is, for example, `MatSeqAIJSetPreallocation()`]
2393: The number of rows and columns must be divisible by blocksize.
2394: This matrix type does not support complex Hermitian operation.
2396: The user MUST specify either the local or global matrix dimensions
2397: (possibly both).
2399: If `PETSC_DECIDE` or `PETSC_DETERMINE` is used for a particular argument on one processor
2400: than it must be used on all processors that share the object for that argument.
2402: If `m` and `n` are not `PETSC_DECIDE`, then the values determines the `PetscLayout` of the matrix and the ranges returned by
2403: `MatGetOwnershipRange()`, `MatGetOwnershipRanges()`, `MatGetOwnershipRangeColumn()`, and `MatGetOwnershipRangesColumn()`.
2405: If the *_nnz parameter is given then the *_nz parameter is ignored
2407: Storage Information:
2408: For a square global matrix we define each processor's diagonal portion
2409: to be its local rows and the corresponding columns (a square submatrix);
2410: each processor's off-diagonal portion encompasses the remainder of the
2411: local matrix (a rectangular submatrix).
2413: The user can specify preallocated storage for the diagonal part of
2414: the local submatrix with either `d_nz` or `d_nnz` (not both). Set
2415: `d_nz` = `PETSC_DEFAULT` and `d_nnz` = `NULL` for PETSc to control dynamic
2416: memory allocation. Likewise, specify preallocated storage for the
2417: off-diagonal part of the local submatrix with `o_nz` or `o_nnz` (not both).
2419: Consider a processor that owns rows 3, 4 and 5 of a parallel matrix. In
2420: the figure below we depict these three local rows and all columns (0-11).
2422: .vb
2423: 0 1 2 3 4 5 6 7 8 9 10 11
2424: --------------------------
2425: row 3 |. . . d d d o o o o o o
2426: row 4 |. . . d d d o o o o o o
2427: row 5 |. . . d d d o o o o o o
2428: --------------------------
2429: .ve
2431: Thus, any entries in the d locations are stored in the d (diagonal)
2432: submatrix, and any entries in the o locations are stored in the
2433: o (off-diagonal) submatrix. Note that the d matrix is stored in
2434: `MATSEQSBAIJ` format and the o submatrix in `MATSEQBAIJ` format.
2436: Now `d_nz` should indicate the number of block nonzeros per row in the upper triangular
2437: plus the diagonal part of the d matrix,
2438: and `o_nz` should indicate the number of block nonzeros per row in the o matrix.
2439: In general, for PDE problems in which most nonzeros are near the diagonal,
2440: one expects `d_nz` >> `o_nz`.
2442: .seealso: [](ch_matrices), `Mat`, `MATSBAIJ`, `MatCreate()`, `MatCreateSeqSBAIJ()`, `MatSetValues()`, `MatCreateBAIJ()`,
2443: `MatGetOwnershipRange()`, `MatGetOwnershipRanges()`, `MatGetOwnershipRangeColumn()`, `MatGetOwnershipRangesColumn()`, `PetscLayout`
2444: @*/
2445: PetscErrorCode MatCreateSBAIJ(MPI_Comm comm, PetscInt bs, PetscInt m, PetscInt n, PetscInt M, PetscInt N, PetscInt d_nz, const PetscInt d_nnz[], PetscInt o_nz, const PetscInt o_nnz[], Mat *A)
2446: {
2447: PetscMPIInt size;
2449: PetscFunctionBegin;
2450: PetscCall(MatCreate(comm, A));
2451: PetscCall(MatSetSizes(*A, m, n, M, N));
2452: PetscCallMPI(MPI_Comm_size(comm, &size));
2453: if (size > 1) {
2454: PetscCall(MatSetType(*A, MATMPISBAIJ));
2455: PetscCall(MatMPISBAIJSetPreallocation(*A, bs, d_nz, d_nnz, o_nz, o_nnz));
2456: } else {
2457: PetscCall(MatSetType(*A, MATSEQSBAIJ));
2458: PetscCall(MatSeqSBAIJSetPreallocation(*A, bs, d_nz, d_nnz));
2459: }
2460: PetscFunctionReturn(PETSC_SUCCESS);
2461: }
2463: static PetscErrorCode MatDuplicate_MPISBAIJ(Mat matin, MatDuplicateOption cpvalues, Mat *newmat)
2464: {
2465: Mat mat;
2466: Mat_MPISBAIJ *a, *oldmat = (Mat_MPISBAIJ *)matin->data;
2467: PetscInt len = 0, nt, bs = matin->rmap->bs, mbs = oldmat->mbs;
2468: PetscScalar *array;
2470: PetscFunctionBegin;
2471: *newmat = NULL;
2473: PetscCall(MatCreate(PetscObjectComm((PetscObject)matin), &mat));
2474: PetscCall(MatSetSizes(mat, matin->rmap->n, matin->cmap->n, matin->rmap->N, matin->cmap->N));
2475: PetscCall(MatSetType(mat, ((PetscObject)matin)->type_name));
2476: PetscCall(PetscLayoutReference(matin->rmap, &mat->rmap));
2477: PetscCall(PetscLayoutReference(matin->cmap, &mat->cmap));
2479: if (matin->hash_active) {
2480: PetscCall(MatSetUp(mat));
2481: } else {
2482: mat->factortype = matin->factortype;
2483: mat->preallocated = PETSC_TRUE;
2484: mat->assembled = PETSC_TRUE;
2485: mat->insertmode = NOT_SET_VALUES;
2487: a = (Mat_MPISBAIJ *)mat->data;
2488: a->bs2 = oldmat->bs2;
2489: a->mbs = oldmat->mbs;
2490: a->nbs = oldmat->nbs;
2491: a->Mbs = oldmat->Mbs;
2492: a->Nbs = oldmat->Nbs;
2494: a->size = oldmat->size;
2495: a->rank = oldmat->rank;
2496: a->donotstash = oldmat->donotstash;
2497: a->roworiented = oldmat->roworiented;
2498: a->rowindices = NULL;
2499: a->rowvalues = NULL;
2500: a->getrowactive = PETSC_FALSE;
2501: a->barray = NULL;
2502: a->rstartbs = oldmat->rstartbs;
2503: a->rendbs = oldmat->rendbs;
2504: a->cstartbs = oldmat->cstartbs;
2505: a->cendbs = oldmat->cendbs;
2507: /* hash table stuff */
2508: a->ht = NULL;
2509: a->hd = NULL;
2510: a->ht_size = 0;
2511: a->ht_flag = oldmat->ht_flag;
2512: a->ht_fact = oldmat->ht_fact;
2513: a->ht_total_ct = 0;
2514: a->ht_insert_ct = 0;
2516: PetscCall(PetscArraycpy(a->rangebs, oldmat->rangebs, a->size + 2));
2517: if (oldmat->colmap) {
2518: #if defined(PETSC_USE_CTABLE)
2519: PetscCall(PetscHMapIDuplicate(oldmat->colmap, &a->colmap));
2520: #else
2521: PetscCall(PetscMalloc1(a->Nbs, &a->colmap));
2522: PetscCall(PetscArraycpy(a->colmap, oldmat->colmap, a->Nbs));
2523: #endif
2524: } else a->colmap = NULL;
2526: if (oldmat->garray && (len = ((Mat_SeqBAIJ *)oldmat->B->data)->nbs)) {
2527: PetscCall(PetscMalloc1(len, &a->garray));
2528: PetscCall(PetscArraycpy(a->garray, oldmat->garray, len));
2529: } else a->garray = NULL;
2531: PetscCall(MatStashCreate_Private(PetscObjectComm((PetscObject)matin), matin->rmap->bs, &mat->bstash));
2532: PetscCall(VecDuplicate(oldmat->lvec, &a->lvec));
2533: PetscCall(VecScatterCopy(oldmat->Mvctx, &a->Mvctx));
2535: PetscCall(VecDuplicate(oldmat->slvec0, &a->slvec0));
2536: PetscCall(VecDuplicate(oldmat->slvec1, &a->slvec1));
2538: PetscCall(VecGetLocalSize(a->slvec1, &nt));
2539: PetscCall(VecGetArray(a->slvec1, &array));
2540: PetscCall(VecCreateSeqWithArray(PETSC_COMM_SELF, 1, bs * mbs, array, &a->slvec1a));
2541: PetscCall(VecCreateSeqWithArray(PETSC_COMM_SELF, 1, nt - bs * mbs, array + bs * mbs, &a->slvec1b));
2542: PetscCall(VecRestoreArray(a->slvec1, &array));
2543: PetscCall(VecGetArray(a->slvec0, &array));
2544: PetscCall(VecCreateSeqWithArray(PETSC_COMM_SELF, 1, nt - bs * mbs, array + bs * mbs, &a->slvec0b));
2545: PetscCall(VecRestoreArray(a->slvec0, &array));
2547: /* ierr = VecScatterCopy(oldmat->sMvctx,&a->sMvctx); - not written yet, replaced by the lazy trick: */
2548: PetscCall(PetscObjectReference((PetscObject)oldmat->sMvctx));
2549: a->sMvctx = oldmat->sMvctx;
2551: PetscCall(MatDuplicate(oldmat->A, cpvalues, &a->A));
2552: PetscCall(MatDuplicate(oldmat->B, cpvalues, &a->B));
2553: }
2554: PetscCall(PetscFunctionListDuplicate(((PetscObject)matin)->qlist, &((PetscObject)mat)->qlist));
2555: *newmat = mat;
2556: PetscFunctionReturn(PETSC_SUCCESS);
2557: }
2559: /* Used for both MPIBAIJ and MPISBAIJ matrices */
2560: #define MatLoad_MPISBAIJ_Binary MatLoad_MPIBAIJ_Binary
2562: static PetscErrorCode MatLoad_MPISBAIJ(Mat mat, PetscViewer viewer)
2563: {
2564: PetscBool isbinary;
2566: PetscFunctionBegin;
2567: PetscCall(PetscObjectTypeCompare((PetscObject)viewer, PETSCVIEWERBINARY, &isbinary));
2568: PetscCheck(isbinary, PetscObjectComm((PetscObject)viewer), PETSC_ERR_SUP, "Viewer type %s not yet supported for reading %s matrices", ((PetscObject)viewer)->type_name, ((PetscObject)mat)->type_name);
2569: PetscCall(MatLoad_MPISBAIJ_Binary(mat, viewer));
2570: PetscFunctionReturn(PETSC_SUCCESS);
2571: }
2573: static PetscErrorCode MatGetRowMaxAbs_MPISBAIJ(Mat A, Vec v, PetscInt idx[])
2574: {
2575: Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data;
2576: Mat_SeqBAIJ *b = (Mat_SeqBAIJ *)a->B->data;
2577: PetscReal atmp;
2578: PetscReal *work, *svalues, *rvalues;
2579: PetscInt i, bs, mbs, *bi, *bj, brow, j, ncols, krow, kcol, col, row, Mbs, bcol;
2580: PetscMPIInt rank, size;
2581: PetscInt *rowners_bs, count, source;
2582: PetscScalar *va;
2583: MatScalar *ba;
2584: MPI_Status stat;
2586: PetscFunctionBegin;
2587: PetscCheck(!idx, PETSC_COMM_SELF, PETSC_ERR_SUP, "Send email to petsc-maint@mcs.anl.gov");
2588: PetscCall(MatGetRowMaxAbs(a->A, v, NULL));
2589: PetscCall(VecGetArray(v, &va));
2591: PetscCallMPI(MPI_Comm_size(PetscObjectComm((PetscObject)A), &size));
2592: PetscCallMPI(MPI_Comm_rank(PetscObjectComm((PetscObject)A), &rank));
2594: bs = A->rmap->bs;
2595: mbs = a->mbs;
2596: Mbs = a->Mbs;
2597: ba = b->a;
2598: bi = b->i;
2599: bj = b->j;
2601: /* find ownerships */
2602: rowners_bs = A->rmap->range;
2604: /* each proc creates an array to be distributed */
2605: PetscCall(PetscCalloc1(bs * Mbs, &work));
2607: /* row_max for B */
2608: if (rank != size - 1) {
2609: for (i = 0; i < mbs; i++) {
2610: ncols = bi[1] - bi[0];
2611: bi++;
2612: brow = bs * i;
2613: for (j = 0; j < ncols; j++) {
2614: bcol = bs * (*bj);
2615: for (kcol = 0; kcol < bs; kcol++) {
2616: col = bcol + kcol; /* local col index */
2617: col += rowners_bs[rank + 1]; /* global col index */
2618: for (krow = 0; krow < bs; krow++) {
2619: atmp = PetscAbsScalar(*ba);
2620: ba++;
2621: row = brow + krow; /* local row index */
2622: if (PetscRealPart(va[row]) < atmp) va[row] = atmp;
2623: if (work[col] < atmp) work[col] = atmp;
2624: }
2625: }
2626: bj++;
2627: }
2628: }
2630: /* send values to its owners */
2631: for (PetscMPIInt dest = rank + 1; dest < size; dest++) {
2632: svalues = work + rowners_bs[dest];
2633: count = rowners_bs[dest + 1] - rowners_bs[dest];
2634: PetscCallMPI(MPIU_Send(svalues, count, MPIU_REAL, dest, rank, PetscObjectComm((PetscObject)A)));
2635: }
2636: }
2638: /* receive values */
2639: if (rank) {
2640: rvalues = work;
2641: count = rowners_bs[rank + 1] - rowners_bs[rank];
2642: for (source = 0; source < rank; source++) {
2643: PetscCallMPI(MPIU_Recv(rvalues, count, MPIU_REAL, MPI_ANY_SOURCE, MPI_ANY_TAG, PetscObjectComm((PetscObject)A), &stat));
2644: /* process values */
2645: for (i = 0; i < count; i++) {
2646: if (PetscRealPart(va[i]) < rvalues[i]) va[i] = rvalues[i];
2647: }
2648: }
2649: }
2651: PetscCall(VecRestoreArray(v, &va));
2652: PetscCall(PetscFree(work));
2653: PetscFunctionReturn(PETSC_SUCCESS);
2654: }
2656: static PetscErrorCode MatSOR_MPISBAIJ(Mat matin, Vec bb, PetscReal omega, MatSORType flag, PetscReal fshift, PetscInt its, PetscInt lits, Vec xx)
2657: {
2658: Mat_MPISBAIJ *mat = (Mat_MPISBAIJ *)matin->data;
2659: PetscInt mbs = mat->mbs, bs = matin->rmap->bs;
2660: PetscScalar *x, *ptr, *from;
2661: Vec bb1;
2662: const PetscScalar *b;
2664: PetscFunctionBegin;
2665: PetscCheck(its > 0 && lits > 0, PETSC_COMM_SELF, PETSC_ERR_ARG_WRONG, "Relaxation requires global its %" PetscInt_FMT " and local its %" PetscInt_FMT " both positive", its, lits);
2666: PetscCheck(bs <= 1, PETSC_COMM_SELF, PETSC_ERR_SUP, "SSOR for block size > 1 is not yet implemented");
2668: if (flag == SOR_APPLY_UPPER) {
2669: PetscCall((*mat->A->ops->sor)(mat->A, bb, omega, flag, fshift, lits, 1, xx));
2670: PetscFunctionReturn(PETSC_SUCCESS);
2671: }
2673: if ((flag & SOR_LOCAL_SYMMETRIC_SWEEP) == SOR_LOCAL_SYMMETRIC_SWEEP) {
2674: if (flag & SOR_ZERO_INITIAL_GUESS) {
2675: PetscCall((*mat->A->ops->sor)(mat->A, bb, omega, flag, fshift, lits, lits, xx));
2676: its--;
2677: }
2679: PetscCall(VecDuplicate(bb, &bb1));
2680: while (its--) {
2681: /* lower triangular part: slvec0b = - B^T*xx */
2682: PetscCall((*mat->B->ops->multtranspose)(mat->B, xx, mat->slvec0b));
2684: /* copy xx into slvec0a */
2685: PetscCall(VecGetArray(mat->slvec0, &ptr));
2686: PetscCall(VecGetArray(xx, &x));
2687: PetscCall(PetscArraycpy(ptr, x, bs * mbs));
2688: PetscCall(VecRestoreArray(mat->slvec0, &ptr));
2690: PetscCall(VecScale(mat->slvec0, -1.0));
2692: /* copy bb into slvec1a */
2693: PetscCall(VecGetArray(mat->slvec1, &ptr));
2694: PetscCall(VecGetArrayRead(bb, &b));
2695: PetscCall(PetscArraycpy(ptr, b, bs * mbs));
2696: PetscCall(VecRestoreArray(mat->slvec1, &ptr));
2698: /* set slvec1b = 0 */
2699: PetscCall(PetscObjectStateIncrease((PetscObject)mat->slvec1b));
2700: PetscCall(VecZeroEntries(mat->slvec1b));
2702: PetscCall(VecScatterBegin(mat->sMvctx, mat->slvec0, mat->slvec1, ADD_VALUES, SCATTER_FORWARD));
2703: PetscCall(VecRestoreArray(xx, &x));
2704: PetscCall(VecRestoreArrayRead(bb, &b));
2705: PetscCall(VecScatterEnd(mat->sMvctx, mat->slvec0, mat->slvec1, ADD_VALUES, SCATTER_FORWARD));
2707: /* upper triangular part: bb1 = bb1 - B*x */
2708: PetscCall((*mat->B->ops->multadd)(mat->B, mat->slvec1b, mat->slvec1a, bb1));
2710: /* local diagonal sweep */
2711: PetscCall((*mat->A->ops->sor)(mat->A, bb1, omega, SOR_SYMMETRIC_SWEEP, fshift, lits, lits, xx));
2712: }
2713: PetscCall(VecDestroy(&bb1));
2714: } else if ((flag & SOR_LOCAL_FORWARD_SWEEP) && (its == 1) && (flag & SOR_ZERO_INITIAL_GUESS)) {
2715: PetscCall((*mat->A->ops->sor)(mat->A, bb, omega, flag, fshift, lits, 1, xx));
2716: } else if ((flag & SOR_LOCAL_BACKWARD_SWEEP) && (its == 1) && (flag & SOR_ZERO_INITIAL_GUESS)) {
2717: PetscCall((*mat->A->ops->sor)(mat->A, bb, omega, flag, fshift, lits, 1, xx));
2718: } else if (flag & SOR_EISENSTAT) {
2719: Vec xx1;
2720: PetscBool hasop;
2721: const PetscScalar *diag;
2722: PetscScalar *sl, scale = (omega - 2.0) / omega;
2723: PetscInt i, n;
2725: if (!mat->xx1) {
2726: PetscCall(VecDuplicate(bb, &mat->xx1));
2727: PetscCall(VecDuplicate(bb, &mat->bb1));
2728: }
2729: xx1 = mat->xx1;
2730: bb1 = mat->bb1;
2732: PetscCall((*mat->A->ops->sor)(mat->A, bb, omega, (MatSORType)(SOR_ZERO_INITIAL_GUESS | SOR_LOCAL_BACKWARD_SWEEP), fshift, lits, 1, xx));
2734: if (!mat->diag) {
2735: /* this is wrong for same matrix with new nonzero values */
2736: PetscCall(MatCreateVecs(matin, &mat->diag, NULL));
2737: PetscCall(MatGetDiagonal(matin, mat->diag));
2738: }
2739: PetscCall(MatHasOperation(matin, MATOP_MULT_DIAGONAL_BLOCK, &hasop));
2741: if (hasop) {
2742: PetscCall(MatMultDiagonalBlock(matin, xx, bb1));
2743: PetscCall(VecAYPX(mat->slvec1a, scale, bb));
2744: } else {
2745: /*
2746: These two lines are replaced by code that may be a bit faster for a good compiler
2747: PetscCall(VecPointwiseMult(mat->slvec1a,mat->diag,xx));
2748: PetscCall(VecAYPX(mat->slvec1a,scale,bb));
2749: */
2750: PetscCall(VecGetArray(mat->slvec1a, &sl));
2751: PetscCall(VecGetArrayRead(mat->diag, &diag));
2752: PetscCall(VecGetArrayRead(bb, &b));
2753: PetscCall(VecGetArray(xx, &x));
2754: PetscCall(VecGetLocalSize(xx, &n));
2755: if (omega == 1.0) {
2756: for (i = 0; i < n; i++) sl[i] = b[i] - diag[i] * x[i];
2757: PetscCall(PetscLogFlops(2.0 * n));
2758: } else {
2759: for (i = 0; i < n; i++) sl[i] = b[i] + scale * diag[i] * x[i];
2760: PetscCall(PetscLogFlops(3.0 * n));
2761: }
2762: PetscCall(VecRestoreArray(mat->slvec1a, &sl));
2763: PetscCall(VecRestoreArrayRead(mat->diag, &diag));
2764: PetscCall(VecRestoreArrayRead(bb, &b));
2765: PetscCall(VecRestoreArray(xx, &x));
2766: }
2768: /* multiply off-diagonal portion of matrix */
2769: PetscCall(PetscObjectStateIncrease((PetscObject)mat->slvec1b));
2770: PetscCall(VecZeroEntries(mat->slvec1b));
2771: PetscCall((*mat->B->ops->multtranspose)(mat->B, xx, mat->slvec0b));
2772: PetscCall(VecGetArray(mat->slvec0, &from));
2773: PetscCall(VecGetArray(xx, &x));
2774: PetscCall(PetscArraycpy(from, x, bs * mbs));
2775: PetscCall(VecRestoreArray(mat->slvec0, &from));
2776: PetscCall(VecRestoreArray(xx, &x));
2777: PetscCall(VecScatterBegin(mat->sMvctx, mat->slvec0, mat->slvec1, ADD_VALUES, SCATTER_FORWARD));
2778: PetscCall(VecScatterEnd(mat->sMvctx, mat->slvec0, mat->slvec1, ADD_VALUES, SCATTER_FORWARD));
2779: PetscCall((*mat->B->ops->multadd)(mat->B, mat->slvec1b, mat->slvec1a, mat->slvec1a));
2781: /* local sweep */
2782: PetscCall((*mat->A->ops->sor)(mat->A, mat->slvec1a, omega, (MatSORType)(SOR_ZERO_INITIAL_GUESS | SOR_LOCAL_FORWARD_SWEEP), fshift, lits, 1, xx1));
2783: PetscCall(VecAXPY(xx, 1.0, xx1));
2784: } else SETERRQ(PETSC_COMM_SELF, PETSC_ERR_SUP, "MatSORType is not supported for SBAIJ matrix format");
2785: PetscFunctionReturn(PETSC_SUCCESS);
2786: }
2788: /*@
2789: MatCreateMPISBAIJWithArrays - creates a `MATMPISBAIJ` matrix using arrays that contain in standard CSR format for the local rows.
2791: Collective
2793: Input Parameters:
2794: + comm - MPI communicator
2795: . bs - the block size, only a block size of 1 is supported
2796: . m - number of local rows (Cannot be `PETSC_DECIDE`)
2797: . n - This value should be the same as the local size used in creating the
2798: x vector for the matrix-vector product $ y = Ax $. (or `PETSC_DECIDE` to have
2799: calculated if `N` is given) For square matrices `n` is almost always `m`.
2800: . M - number of global rows (or `PETSC_DETERMINE` to have calculated if `m` is given)
2801: . N - number of global columns (or `PETSC_DETERMINE` to have calculated if `n` is given)
2802: . i - row indices; that is i[0] = 0, i[row] = i[row-1] + number of block elements in that row block row of the matrix
2803: . j - column indices
2804: - a - matrix values
2806: Output Parameter:
2807: . mat - the matrix
2809: Level: intermediate
2811: Notes:
2812: The `i`, `j`, and `a` arrays ARE copied by this routine into the internal format used by PETSc;
2813: thus you CANNOT change the matrix entries by changing the values of `a` after you have
2814: called this routine. Use `MatCreateMPIAIJWithSplitArrays()` to avoid needing to copy the arrays.
2816: The `i` and `j` indices are 0 based, and `i` indices are indices corresponding to the local `j` array.
2818: .seealso: [](ch_matrices), `Mat`, `MATMPISBAIJ`, `MatCreate()`, `MatCreateSeqAIJ()`, `MatSetValues()`, `MatMPIAIJSetPreallocation()`, `MatMPIAIJSetPreallocationCSR()`,
2819: `MATMPIAIJ`, `MatCreateAIJ()`, `MatCreateMPIAIJWithSplitArrays()`, `MatMPISBAIJSetPreallocationCSR()`
2820: @*/
2821: PetscErrorCode MatCreateMPISBAIJWithArrays(MPI_Comm comm, PetscInt bs, PetscInt m, PetscInt n, PetscInt M, PetscInt N, const PetscInt i[], const PetscInt j[], const PetscScalar a[], Mat *mat)
2822: {
2823: PetscFunctionBegin;
2824: PetscCheck(!i[0], PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "i (row indices) must start with 0");
2825: PetscCheck(m >= 0, PETSC_COMM_SELF, PETSC_ERR_ARG_OUTOFRANGE, "local number of rows (m) cannot be PETSC_DECIDE, or negative");
2826: PetscCall(MatCreate(comm, mat));
2827: PetscCall(MatSetSizes(*mat, m, n, M, N));
2828: PetscCall(MatSetType(*mat, MATMPISBAIJ));
2829: PetscCall(MatMPISBAIJSetPreallocationCSR(*mat, bs, i, j, a));
2830: PetscFunctionReturn(PETSC_SUCCESS);
2831: }
2833: /*@
2834: MatMPISBAIJSetPreallocationCSR - Creates a sparse parallel matrix in `MATMPISBAIJ` format using the given nonzero structure and (optional) numerical values
2836: Collective
2838: Input Parameters:
2839: + B - the matrix
2840: . bs - the block size
2841: . i - the indices into `j` for the start of each local row (indices start with zero)
2842: . j - the column indices for each local row (indices start with zero) these must be sorted for each row
2843: - v - optional values in the matrix, pass `NULL` if not provided
2845: Level: advanced
2847: Notes:
2848: The `i`, `j`, and `v` arrays ARE copied by this routine into the internal format used by PETSc;
2849: thus you CANNOT change the matrix entries by changing the values of `v` after you have
2850: called this routine.
2852: Though this routine has Preallocation() in the name it also sets the exact nonzero locations of the matrix entries
2853: and usually the numerical values as well
2855: Any entries passed in that are below the diagonal are ignored
2857: .seealso: [](ch_matrices), `Mat`, `MATMPISBAIJ`, `MatCreate()`, `MatCreateSeqAIJ()`, `MatSetValues()`, `MatMPIBAIJSetPreallocation()`, `MatCreateAIJ()`, `MATMPIAIJ`,
2858: `MatCreateMPISBAIJWithArrays()`
2859: @*/
2860: PetscErrorCode MatMPISBAIJSetPreallocationCSR(Mat B, PetscInt bs, const PetscInt i[], const PetscInt j[], const PetscScalar v[])
2861: {
2862: PetscFunctionBegin;
2863: PetscTryMethod(B, "MatMPISBAIJSetPreallocationCSR_C", (Mat, PetscInt, const PetscInt[], const PetscInt[], const PetscScalar[]), (B, bs, i, j, v));
2864: PetscFunctionReturn(PETSC_SUCCESS);
2865: }
2867: PetscErrorCode MatCreateMPIMatConcatenateSeqMat_MPISBAIJ(MPI_Comm comm, Mat inmat, PetscInt n, MatReuse scall, Mat *outmat)
2868: {
2869: PetscInt m, N, i, rstart, nnz, Ii, bs, cbs;
2870: PetscInt *indx;
2871: PetscScalar *values;
2873: PetscFunctionBegin;
2874: PetscCall(MatGetSize(inmat, &m, &N));
2875: if (scall == MAT_INITIAL_MATRIX) { /* symbolic phase */
2876: Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ *)inmat->data;
2877: PetscInt *dnz, *onz, mbs, Nbs, nbs;
2878: PetscInt *bindx, rmax = a->rmax, j;
2879: PetscMPIInt rank, size;
2881: PetscCall(MatGetBlockSizes(inmat, &bs, &cbs));
2882: mbs = m / bs;
2883: Nbs = N / cbs;
2884: if (n == PETSC_DECIDE) PetscCall(PetscSplitOwnershipBlock(comm, cbs, &n, &N));
2885: nbs = n / cbs;
2887: PetscCall(PetscMalloc1(rmax, &bindx));
2888: MatPreallocateBegin(comm, mbs, nbs, dnz, onz); /* inline function, output __end and __rstart are used below */
2890: PetscCallMPI(MPI_Comm_rank(comm, &rank));
2891: PetscCallMPI(MPI_Comm_rank(comm, &size));
2892: if (rank == size - 1) {
2893: /* Check sum(nbs) = Nbs */
2894: PetscCheck(__end == Nbs, PETSC_COMM_SELF, PETSC_ERR_ARG_INCOMP, "Sum of local block columns %" PetscInt_FMT " != global block columns %" PetscInt_FMT, __end, Nbs);
2895: }
2897: rstart = __rstart; /* block rstart of *outmat; see inline function MatPreallocateBegin */
2898: PetscCall(MatSetOption(inmat, MAT_GETROW_UPPERTRIANGULAR, PETSC_TRUE));
2899: for (i = 0; i < mbs; i++) {
2900: PetscCall(MatGetRow_SeqSBAIJ(inmat, i * bs, &nnz, &indx, NULL)); /* non-blocked nnz and indx */
2901: nnz = nnz / bs;
2902: for (j = 0; j < nnz; j++) bindx[j] = indx[j * bs] / bs;
2903: PetscCall(MatPreallocateSet(i + rstart, nnz, bindx, dnz, onz));
2904: PetscCall(MatRestoreRow_SeqSBAIJ(inmat, i * bs, &nnz, &indx, NULL));
2905: }
2906: PetscCall(MatSetOption(inmat, MAT_GETROW_UPPERTRIANGULAR, PETSC_FALSE));
2907: PetscCall(PetscFree(bindx));
2909: PetscCall(MatCreate(comm, outmat));
2910: PetscCall(MatSetSizes(*outmat, m, n, PETSC_DETERMINE, PETSC_DETERMINE));
2911: PetscCall(MatSetBlockSizes(*outmat, bs, cbs));
2912: PetscCall(MatSetType(*outmat, MATSBAIJ));
2913: PetscCall(MatSeqSBAIJSetPreallocation(*outmat, bs, 0, dnz));
2914: PetscCall(MatMPISBAIJSetPreallocation(*outmat, bs, 0, dnz, 0, onz));
2915: MatPreallocateEnd(dnz, onz);
2916: }
2918: /* numeric phase */
2919: PetscCall(MatGetBlockSizes(inmat, &bs, &cbs));
2920: PetscCall(MatGetOwnershipRange(*outmat, &rstart, NULL));
2922: PetscCall(MatSetOption(inmat, MAT_GETROW_UPPERTRIANGULAR, PETSC_TRUE));
2923: for (i = 0; i < m; i++) {
2924: PetscCall(MatGetRow_SeqSBAIJ(inmat, i, &nnz, &indx, &values));
2925: Ii = i + rstart;
2926: PetscCall(MatSetValues(*outmat, 1, &Ii, nnz, indx, values, INSERT_VALUES));
2927: PetscCall(MatRestoreRow_SeqSBAIJ(inmat, i, &nnz, &indx, &values));
2928: }
2929: PetscCall(MatSetOption(inmat, MAT_GETROW_UPPERTRIANGULAR, PETSC_FALSE));
2930: PetscCall(MatAssemblyBegin(*outmat, MAT_FINAL_ASSEMBLY));
2931: PetscCall(MatAssemblyEnd(*outmat, MAT_FINAL_ASSEMBLY));
2932: PetscFunctionReturn(PETSC_SUCCESS);
2933: }